a lot of simplification finally. all works

llama_stack/core/server/server.py

@@ -232,14 +232,25 @@ def create_dynamic_typed_route(func: Any, method: str, route: str) -> Callable:
 
         await log_request_pre_validation(request)
 
+        test_context_token = None
+
         # Use context manager with both provider data and auth attributes
         with request_provider_data_context(request.headers, user):
+            if os.environ.get("LLAMA_STACK_TEST_INFERENCE_MODE"):
+                from llama_stack.core.testing_context import (
+                    TEST_CONTEXT,
+                    reset_test_context,
+                    sync_test_context_from_provider_data,
+                )
+
+                test_context_token = sync_test_context_from_provider_data()
+
             is_streaming = is_streaming_request(func.__name__, request, **kwargs)
 
             try:
                 if is_streaming:
                     gen = preserve_contexts_async_generator(
-                        sse_generator(func(**kwargs)), [CURRENT_TRACE_CONTEXT, PROVIDER_DATA_VAR]
+                        sse_generator(func(**kwargs)), [CURRENT_TRACE_CONTEXT, PROVIDER_DATA_VAR, TEST_CONTEXT]
                     )
                     return StreamingResponse(gen, media_type="text/event-stream")
                 else:
@@ -258,6 +269,9 @@ def create_dynamic_typed_route(func: Any, method: str, route: str) -> Callable:
                 else:
                     logger.error(f"Error executing endpoint {route=} {method=}: {str(e)}")
                 raise translate_exception(e) from e
+            finally:
+                if test_context_token is not None:
+                    reset_test_context(test_context_token)
 
     sig = inspect.signature(func)
 

llama_stack/core/testing_context.py

@@ -0,0 +1,44 @@
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the terms described in the LICENSE file in
+# the root directory of this source tree.
+
+import os
+from contextvars import ContextVar
+
+from llama_stack.core.request_headers import PROVIDER_DATA_VAR
+
+TEST_CONTEXT: ContextVar[str | None] = ContextVar("llama_stack_test_context", default=None)
+
+
+def get_test_context() -> str | None:
+    return TEST_CONTEXT.get()
+
+
+def set_test_context(value: str | None):
+    return TEST_CONTEXT.set(value)
+
+
+def reset_test_context(token) -> None:
+    TEST_CONTEXT.reset(token)
+
+
+def sync_test_context_from_provider_data():
+    """Sync test context from provider data when running in server test mode."""
+    if "LLAMA_STACK_TEST_INFERENCE_MODE" not in os.environ:
+        return None
+
+    stack_config_type = os.environ.get("LLAMA_STACK_TEST_STACK_CONFIG_TYPE", "library_client")
+    if stack_config_type != "server":
+        return None
+
+    try:
+        provider_data = PROVIDER_DATA_VAR.get()
+    except LookupError:
+        provider_data = None
+
+    if provider_data and "__test_id" in provider_data:
+        return TEST_CONTEXT.set(provider_data["__test_id"])
+
+    return None

llama_stack/testing/api_recorder.py

@@ -9,7 +9,6 @@ from __future__ import annotations  # for forward references
 import hashlib
 import json
 import os
-import re
 from collections.abc import Callable, Generator
 from contextlib import contextmanager
 from enum import StrEnum
@@ -35,12 +34,10 @@ _original_methods: dict[str, Any] = {}
 _id_counters: dict[str, dict[str, int]] = {}
 
 # Test context uses ContextVar since it changes per-test and needs async isolation
-from contextvars import ContextVar
-
-_test_context: ContextVar[str | None] = ContextVar("_test_context", default=None)
-
 from openai.types.completion_choice import CompletionChoice
 
+from llama_stack.core.testing_context import get_test_context
+
 # update the "finish_reason" field, since its type definition is wrong (no None is accepted)
 CompletionChoice.model_fields["finish_reason"].annotation = Literal["stop", "length", "content_filter"] | None
 CompletionChoice.model_rebuild()
@@ -60,6 +57,7 @@ _ID_KIND_PREFIXES: dict[str, str] = {
     "file": "file-",
     "vector_store": "vs_",
     "vector_store_file_batch": "batch_",
+    "tool_call": "call_",
 }
 
 
@@ -68,176 +66,32 @@ def _allocate_test_scoped_id(kind: str) -> str | None:
 
     global _id_counters
 
-    test_id = _test_context.get()
+    test_id = get_test_context()
     prefix = _ID_KIND_PREFIXES.get(kind)
 
     if prefix is None:
         return None
 
-    key = test_id or "__global__"
+    if not test_id:
+        raise ValueError(f"Test ID is required for {kind} ID allocation")
 
+    key = test_id
     if key not in _id_counters:
         _id_counters[key] = {}
 
-    counter = _id_counters[key].get(kind, 0) + 1
+    # each test should get a contiguous block of IDs otherwise we will get
+    # collisions between tests inside other systems (like file storage) which
+    # expect IDs to be unique
+    test_hash = hashlib.sha256(test_id.encode()).hexdigest()
+    test_hash_int = int(test_hash, 16)
+    counter = test_hash_int % 1000000000000
+
+    counter = _id_counters[key].get(kind, counter) + 1
     _id_counters[key][kind] = counter
 
     return f"{prefix}{counter}"
 
 
-class _IdCanonicalizer:
-    PATTERN = re.compile(r"(file-[A-Za-z0-9_-]+|vs_[A-Za-z0-9_-]+|batch_[A-Za-z0-9_-]+)")
-
-    def __init__(self) -> None:
-        self._mappings: dict[str, dict[str, str]] = {kind: {} for kind in _ID_KIND_PREFIXES}
-        self._counters: dict[str, int] = dict.fromkeys(_ID_KIND_PREFIXES, 0)
-
-    def canonicalize(self, obj: Any) -> Any:
-        if isinstance(obj, dict):
-            return {k: self._canonicalize_value(k, v) for k, v in obj.items()}
-        if isinstance(obj, list):
-            return [self.canonicalize(item) for item in obj]
-        if isinstance(obj, str):
-            return self._canonicalize_string(obj)
-        return obj
-
-    def _canonicalize_value(self, key: str, value: Any) -> Any:
-        if key in {"vector_db_id", "vector_store_id", "bank_id"} and isinstance(value, str):
-            return self._canonicalize_string(value)
-        if key == "document_id" and isinstance(value, str) and value.startswith("file-"):
-            return self._canonicalize_string(value)
-        return self.canonicalize(value)
-
-    def _canonicalize_string(self, value: str) -> str:
-        def replace(match: re.Match[str]) -> str:
-            token = match.group(0)
-            if token.startswith("file-"):
-                return self._mapped_value("file", token)
-            if token.startswith("vs_"):
-                return self._mapped_value("vector_store", token)
-            if token.startswith("batch_"):
-                return self._mapped_value("vector_store_file_batch", token)
-            return token
-
-        return self.PATTERN.sub(replace, value)
-
-    def _mapped_value(self, kind: str, original: str) -> str:
-        mapping = self._mappings[kind]
-        if original not in mapping:
-            self._counters[kind] += 1
-            mapping[original] = f"{_ID_KIND_PREFIXES[kind]}{self._counters[kind]}"
-        return mapping[original]
-
-
-def _canonicalize_for_hashing(obj: Any) -> Any:
-    canonicalizer = _IdCanonicalizer()
-    return canonicalizer.canonicalize(obj)
-
-
-def _chunk_text_content(chunk: Any) -> tuple[str | None, bool]:
-    """Return (content, has_structured_fields) for OpenAI chat completion chunks."""
-    choices = getattr(chunk, "choices", None)
-    if not choices:
-        return None, False
-
-    delta = choices[0].delta
-    content = getattr(delta, "content", None)
-    if not content:
-        return None, False
-
-    has_structured = bool(getattr(delta, "tool_calls", None) or getattr(delta, "function_call", None))
-    return content, has_structured
-
-
-def _chunk_with_content(chunk: Any, content: str) -> Any:
-    """Return a copy of the chunk with delta.content replaced by the provided string."""
-    choices = getattr(chunk, "choices", None)
-    if not choices:
-        return chunk
-
-    updated_choices = []
-    for choice in choices:
-        delta = choice.delta
-        if getattr(delta, "content", None) is not None:
-            new_delta = delta.model_copy(update={"content": content})
-            updated_choices.append(choice.model_copy(update={"delta": new_delta}))
-        else:
-            updated_choices.append(choice)
-
-    return chunk.model_copy(update={"choices": updated_choices})
-
-
-def _ends_with_partial_identifier(text: str) -> bool:
-    """Return True if text ends in an incomplete file identifier."""
-    match = re.search(r"(?:<\|)?file-[A-Za-z0-9_-]*$", text)
-    if not match:
-        return False
-
-    token = match.group()
-    enclosed = token.startswith("<|")
-    if enclosed and not token.endswith("|>"):
-        return True
-
-    if enclosed:
-        core = token[2:-2] if token.endswith("|>") else token[2:]
-    else:
-        core = token
-
-    suffix = core[len("file-") :]
-    if len(suffix) < 16:
-        return True
-    if not re.fullmatch(r"[A-Za-z0-9_-]+", suffix):
-        return True
-
-    return False
-
-
-def _has_safe_boundary(text: str) -> bool:
-    if not text:
-        return False
-
-    last_char = text[-1]
-    if last_char.isspace():
-        return True
-
-    return last_char in ".,?!;:)]}>\"'"
-
-
-def _coalesce_streaming_chunks(chunks: list[Any]) -> list[Any]:
-    """Merge adjacent text chunks to avoid breaking identifiers across boundaries."""
-    result: list[Any] = []
-    pending_chunk: Any | None = None
-    pending_content = ""
-
-    for chunk in chunks:
-        content, has_structured = _chunk_text_content(chunk)
-
-        if content is None or has_structured:
-            if pending_chunk is not None:
-                result.append(_chunk_with_content(pending_chunk, pending_content))
-                pending_chunk = None
-                pending_content = ""
-
-            result.append(chunk)
-            continue
-
-        if pending_chunk is None:
-            pending_chunk = chunk
-            pending_content = content
-        else:
-            pending_content += content
-
-        if (not _ends_with_partial_identifier(pending_content)) and _has_safe_boundary(pending_content):
-            result.append(_chunk_with_content(pending_chunk, pending_content))
-            pending_chunk = None
-            pending_content = ""
-
-    if pending_chunk is not None:
-        result.append(_chunk_with_content(pending_chunk, pending_content))
-
-    return result
-
-
 def _deterministic_id_override(kind: str, factory: Callable[[], str]) -> str:
     deterministic_id = _allocate_test_scoped_id(kind)
     if deterministic_id is not None:
@@ -262,12 +116,12 @@ def normalize_inference_request(method: str, url: str, headers: dict[str, Any],
     normalized: dict[str, Any] = {
         "method": method.upper(),
         "endpoint": parsed.path,
-        "body": _canonicalize_for_hashing(body),
+        "body": body,
     }
 
     # Include test_id for isolation, except for shared infrastructure endpoints
     if parsed.path not in ("/api/tags", "/v1/models"):
-        normalized["test_id"] = _test_context.get()
+        normalized["test_id"] = get_test_context()
 
     # Create hash - sort_keys=True ensures deterministic ordering
     normalized_json = json.dumps(normalized, sort_keys=True)
@@ -279,7 +133,7 @@ def normalize_tool_request(provider_name: str, tool_name: str, kwargs: dict[str,
     normalized = {
         "provider": provider_name,
         "tool_name": tool_name,
-        "kwargs": _canonicalize_for_hashing(kwargs),
+        "kwargs": kwargs,
     }
 
     # Create hash - sort_keys=True ensures deterministic ordering
@@ -287,33 +141,6 @@ def normalize_tool_request(provider_name: str, tool_name: str, kwargs: dict[str,
     return hashlib.sha256(normalized_json.encode()).hexdigest()
 
 
-def _sync_test_context_from_provider_data():
-    """In server mode, sync test ID from provider_data to _test_context.
-
-    This ensures that storage operations (which read from _test_context) work correctly
-    in server mode where the test ID arrives via HTTP header → provider_data.
-
-    Returns a token to reset _test_context, or None if no sync was needed.
-    """
-    stack_config_type = os.environ.get("LLAMA_STACK_TEST_STACK_CONFIG_TYPE", "library_client")
-
-    if stack_config_type != "server":
-        return None
-
-    try:
-        from llama_stack.core.request_headers import PROVIDER_DATA_VAR
-
-        provider_data = PROVIDER_DATA_VAR.get()
-
-        if provider_data and "__test_id" in provider_data:
-            test_id = provider_data["__test_id"]
-            return _test_context.set(test_id)
-    except ImportError:
-        pass
-
-    return None
-
-
 def patch_httpx_for_test_id():
     """Patch client _prepare_request methods to inject test ID into provider data header.
 
@@ -335,13 +162,12 @@ def patch_httpx_for_test_id():
     def patched_prepare_request(self, request):
         # Call original first (it's a sync method that returns None)
         # Determine which original to call based on client type
-        if "llama_stack_client" in self.__class__.__module__:
         _original_methods["llama_stack_client_prepare_request"](self, request)
         _original_methods["openai_prepare_request"](self, request)
 
         # Only inject test ID in server mode
         stack_config_type = os.environ.get("LLAMA_STACK_TEST_STACK_CONFIG_TYPE", "library_client")
-        test_id = _test_context.get()
+        test_id = get_test_context()
 
         if stack_config_type == "server" and test_id:
             provider_data_header = request.headers.get("X-LlamaStack-Provider-Data")
@@ -482,8 +308,6 @@ class ResponseStorage:
     def __init__(self, base_dir: Path):
         self.base_dir = base_dir
         # Don't create responses_dir here - determine it per-test at runtime
-        self._legacy_index: dict[str, Path] = {}
-        self._scanned_dirs: set[Path] = set()
 
     def _get_test_dir(self) -> Path:
         """Get the recordings directory in the test file's parent directory.
@@ -491,7 +315,7 @@ class ResponseStorage:
         For test at "tests/integration/inference/test_foo.py::test_bar",
         returns "tests/integration/inference/recordings/".
         """
-        test_id = _test_context.get()
+        test_id = get_test_context()
         if test_id:
             # Extract the directory path from the test nodeid
             # e.g., "tests/integration/inference/test_basic.py::test_foo[params]"
@@ -506,7 +330,7 @@ class ResponseStorage:
             # Fallback for non-test contexts
             return self.base_dir / "recordings"
 
-    def _ensure_directories(self):
+    def _ensure_directory(self):
         """Ensure test-specific directories exist."""
         test_dir = self._get_test_dir()
         test_dir.mkdir(parents=True, exist_ok=True)
@@ -514,7 +338,7 @@ class ResponseStorage:
 
     def store_recording(self, request_hash: str, request: dict[str, Any], response: dict[str, Any]):
         """Store a request/response pair."""
-        responses_dir = self._ensure_directories()
+        responses_dir = self._ensure_directory()
 
         # Use FULL hash (not truncated)
         response_file = f"{request_hash}.json"
@@ -543,7 +367,7 @@ class ResponseStorage:
         with open(response_path, "w") as f:
             json.dump(
                 {
-                    "test_id": _test_context.get(),
+                    "test_id": get_test_context(),
                     "request": request,
                     "response": serialized_response,
                     "id_normalization_mapping": {},
@@ -554,8 +378,6 @@ class ResponseStorage:
             f.write("\n")
             f.flush()
 
-        self._legacy_index[request_hash] = response_path
-
     def find_recording(self, request_hash: str) -> dict[str, Any] | None:
         """Find a recorded response by request hash.
 
@@ -579,52 +401,6 @@ class ResponseStorage:
         if fallback_path.exists():
             return _recording_from_file(fallback_path)
 
-        return self._find_in_legacy_index(request_hash, [test_dir, fallback_dir])
-
-    def _find_in_legacy_index(self, request_hash: str, directories: list[Path]) -> dict[str, Any] | None:
-        for directory in directories:
-            if not directory.exists() or directory in self._scanned_dirs:
-                continue
-
-            for path in directory.glob("*.json"):
-                try:
-                    with open(path) as f:
-                        data = json.load(f)
-                except Exception:
-                    continue
-
-                request = data.get("request")
-                if not request:
-                    continue
-
-                body = request.get("body")
-                canonical_body = _canonicalize_for_hashing(body) if isinstance(body, dict | list) else body
-
-                token = None
-                test_id = data.get("test_id")
-                if test_id:
-                    token = _test_context.set(test_id)
-
-                try:
-                    legacy_hash = normalize_inference_request(
-                        request.get("method", ""),
-                        request.get("url", ""),
-                        request.get("headers", {}),
-                        canonical_body,
-                    )
-                finally:
-                    if token is not None:
-                        _test_context.reset(token)
-
-                if legacy_hash not in self._legacy_index:
-                    self._legacy_index[legacy_hash] = path
-
-            self._scanned_dirs.add(directory)
-
-        legacy_path = self._legacy_index.get(request_hash)
-        if legacy_path and legacy_path.exists():
-            return _recording_from_file(legacy_path)
-
         return None
 
     def _model_list_responses(self, request_hash: str) -> list[dict[str, Any]]:
@@ -740,10 +516,6 @@ async def _patched_tool_invoke_method(
         # Normal operation
         return await original_method(self, tool_name, kwargs)
 
-    # In server mode, sync test ID from provider_data to _test_context for storage operations
-    test_context_token = _sync_test_context_from_provider_data()
-
-    try:
     request_hash = normalize_tool_request(provider_name, tool_name, kwargs)
 
     if _current_mode in (APIRecordingMode.REPLAY, APIRecordingMode.RECORD_IF_MISSING):
@@ -763,7 +535,7 @@ async def _patched_tool_invoke_method(
         result = await original_method(self, tool_name, kwargs)
 
         request_data = {
-                "test_id": _test_context.get(),
+            "test_id": get_test_context(),
             "provider": provider_name,
             "tool_name": tool_name,
             "kwargs": kwargs,
@@ -776,10 +548,6 @@ async def _patched_tool_invoke_method(
 
     else:
         raise AssertionError(f"Invalid mode: {_current_mode}")
-    finally:
-        # Reset test context if we set it in server mode
-        if test_context_token is not None:
-            _test_context.reset(test_context_token)
 
 
 async def _patched_inference_method(original_method, self, client_type, endpoint, *args, **kwargs):
@@ -794,10 +562,6 @@ async def _patched_inference_method(original_method, self, client_type, endpoint
         else:
             return await original_method(self, *args, **kwargs)
 
-    # In server mode, sync test ID from provider_data to _test_context for storage operations
-    test_context_token = _sync_test_context_from_provider_data()
-
-    try:
     # Get base URL based on client type
     if client_type == "openai":
         base_url = str(self._client.base_url)
@@ -836,9 +600,6 @@ async def _patched_inference_method(original_method, self, client_type, endpoint
         if recording:
             response_body = recording["response"]["body"]
 
-                if recording["response"].get("is_streaming", False) and isinstance(response_body, list):
-                    response_body = _coalesce_streaming_chunks(response_body)
-
             if recording["response"].get("is_streaming", False):
 
                 async def replay_stream():
@@ -882,11 +643,9 @@ async def _patched_inference_method(original_method, self, client_type, endpoint
         if is_streaming:
             # For streaming responses, we need to collect all chunks immediately before yielding
             # This ensures the recording is saved even if the generator isn't fully consumed
-                raw_chunks: list[Any] = []
+            chunks: list[Any] = []
             async for chunk in response:
-                    raw_chunks.append(chunk)
+                chunks.append(chunk)
-
-                chunks = _coalesce_streaming_chunks(raw_chunks)
 
             # Store the recording immediately
             response_data = {"body": chunks, "is_streaming": True}
@@ -905,9 +664,6 @@ async def _patched_inference_method(original_method, self, client_type, endpoint
 
     else:
         raise AssertionError(f"Invalid mode: {mode}")
-    finally:
-        if test_context_token:
-            _test_context.reset(test_context_token)
 
 
 def patch_inference_clients():

tests/integration/common/recordings/224f7e7bd332f0ada326039866d13b7f167d5bfa91ce752022010f1e885d869f.json

@@ -0,0 +1,107 @@
+{
+  "test_id": null,
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Which planet has rings around it with a name starting with letter S?"
+        }
+      ],
+      "stream": true
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-224f7e7bd332",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f33640a400",
+          "usage": null,
+          "obfuscation": "QgpggKqlsUi"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-224f7e7bd332",
+          "choices": [
+            {
+              "delta": {
+                "content": "The planet with rings around it that starts with the letter S is Saturn.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f33640a400",
+          "usage": null,
+          "obfuscation": "HYFma3xA0U"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-224f7e7bd332",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f33640a400",
+          "usage": null,
+          "obfuscation": "GqfSkdB"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/common/recordings/a4416482053457914b5834398c2664ceb843d8c7deaec80a59d5e20dbb1ca090.json

@@ -0,0 +1,107 @@
+{
+  "test_id": null,
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Which planet do humans live on?"
+        }
+      ],
+      "stream": true
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-a44164820534",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_1827dd0c55",
+          "usage": null,
+          "obfuscation": "Uk6yP9DR13H"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-a44164820534",
+          "choices": [
+            {
+              "delta": {
+                "content": "Humans live on Earth.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_1827dd0c55",
+          "usage": null,
+          "obfuscation": "6RKlE6NFLf"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-a44164820534",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_1827dd0c55",
+          "usage": null,
+          "obfuscation": "IhlSKzm"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/common/recordings/a6ad8748dce1ebe53352c6ac4ccd9b209d614ce5c6ff86992b4aed3dc344eafc.json

@@ -0,0 +1,124 @@
+{
+  "test_id": null,
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": [
+            {
+              "type": "text",
+              "text": "what teams are playing in this image?"
+            }
+          ]
+        },
+        {
+          "role": "user",
+          "content": [
+            {
+              "type": "image_url",
+              "image_url": {
+                "url": "https://upload.wikimedia.org/wikipedia/commons/3/3b/LeBron_James_Layup_%28Cleveland_vs_Brooklyn_2018%29.jpg",
+                "detail": "auto"
+              }
+            }
+          ]
+        }
+      ],
+      "stream": true
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-a6ad8748dce1",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "XomWZpEB3cK"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-a6ad8748dce1",
+          "choices": [
+            {
+              "delta": {
+                "content": "The teams playing in the image are the Cleveland Cavaliers and the Brooklyn Nets.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "w3SqawrKwS"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-a6ad8748dce1",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "kkQyIuu"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/common/recordings/models-64a2277c90f0f42576f60c1030e3a020403d34a95f56931b792d5939f4cebc57-4c45d25f.json

@@ -849,5 +849,6 @@
       }
     ],
     "is_streaming": false
-  }
+  },
+  "id_normalization_mapping": {}
 }

tests/integration/conftest.py

@@ -59,15 +59,13 @@ def _track_test_context(request):
     This fixture runs for every test and stores the test's nodeid in a contextvar
     that the recording system can access to determine which subdirectory to use.
     """
-    from llama_stack.testing.api_recorder import _test_context
+    from llama_stack.core.testing_context import reset_test_context, set_test_context
 
-    # Store the test nodeid (e.g., "tests/integration/responses/test_basic.py::test_foo[params]")
+    token = set_test_context(request.node.nodeid)
-    token = _test_context.set(request.node.nodeid)
 
     yield
 
-    # Cleanup
+    reset_test_context(token)
-    _test_context.reset(token)
 
 
 def pytest_runtest_teardown(item):

tests/integration/responses/fixtures/test_cases.py

@@ -41,32 +41,33 @@ basic_test_cases = [
         ),
         id="saturn",
     ),
-    pytest.param(
+    # TODO: Add image input test case, since this test case got Nerfed by OpenAI with a refusal
-        ResponsesTestCase(
+    # pytest.param(
-            input=[
+    #     ResponsesTestCase(
-                {
+    #         input=[
-                    "role": "user",
+    #             {
-                    "content": [
+    #                 "role": "user",
-                        {
+    #                 "content": [
-                            "type": "input_text",
+    #                     {
-                            "text": "what teams are playing in this image?",
+    #                         "type": "input_text",
-                        }
+    #                         "text": "what teams are playing in this image?",
-                    ],
+    #                     }
-                },
+    #                 ],
-                {
+    #             },
-                    "role": "user",
+    #             {
-                    "content": [
+    #                 "role": "user",
-                        {
+    #                 "content": [
-                            "type": "input_image",
+    #                     {
-                            "image_url": "https://upload.wikimedia.org/wikipedia/commons/3/3b/LeBron_James_Layup_%28Cleveland_vs_Brooklyn_2018%29.jpg",
+    #                         "type": "input_image",
-                        }
+    #                         "image_url": "https://upload.wikimedia.org/wikipedia/commons/3/3b/LeBron_James_Layup_%28Cleveland_vs_Brooklyn_2018%29.jpg",
-                    ],
+    #                     }
-                },
+    #                 ],
-            ],
+    #             },
-            expected="brooklyn nets",
+    #         ],
-        ),
+    #         expected="brooklyn nets",
-        id="image_input",
+    #     ),
-    ),
+    #     id="image_input",
+    # ),
 ]
 
 # Multi-turn test cases

tests/integration/responses/recordings/07b6475c42139e6e3f06323eeef10cba663a61796f4667367aa5084f2e2c2eb6.json

@@ -0,0 +1,372 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_sequential_file_search[openai_client-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_aPe1vS1v5bIwPgl789D5bfmW",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_aPe1vS1v5bIwPgl789D5bfmW",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-690481987689, score: 2.5781234969335522, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-690481987689', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-690481987689|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        },
+        {
+          "role": "assistant",
+          "content": "The Llama 4 Maverick model has 128 experts in its mixture of experts architecture <|file-690481987689|>."
+        },
+        {
+          "role": "user",
+          "content": "Can you tell me more about the architecture?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_UqyVZyvMh30eQuiKg7lJIUhQ",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model architecture\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_UqyVZyvMh30eQuiKg7lJIUhQ",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-690481987689, score: 1.9327567816402336, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-690481987689', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-690481987689|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model architecture\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "MR85AdN9cL5"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": "The Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "PU1MvlGgbp"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model's architecture utilizes a mixture of experts,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "j5k5TwcdwAGF"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": " comprising ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "RY"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts <|file-690481987689|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "zfJzUoSQia"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": " Unfortunately,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "myxQ8DZiq0bzUFn"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": " the retrieved data did not provide more detailed information beyond this specific aspect of the architecture.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "tqzrfbJ9V"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": " Would you like me to search for additional details?",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "p8JOT5P"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-07b6475c4213",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "tA2QeAL"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/0fd12925d27e6824a956559c79b4c90723b11660134454284bddc28e2275f770.json

@@ -0,0 +1,219 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_sequential_file_search[openai_client-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_aPe1vS1v5bIwPgl789D5bfmW",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_aPe1vS1v5bIwPgl789D5bfmW",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-690481987689, score: 2.5781234969335522, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-690481987689', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-690481987689|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-0fd12925d27e",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "SgS7tknb3vK"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-0fd12925d27e",
+          "choices": [
+            {
+              "delta": {
+                "content": "The Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Rf8yvaq0Sm"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-0fd12925d27e",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model has ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "KwthQ5gO2NxL"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-0fd12925d27e",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts in its mixture of experts architecture <|file-690481987689|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "wmUFlFBAXg"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-0fd12925d27e",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "El7QNNB"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/127a97b42f235e9a22414005f5c941476af8e8a6a211898394fee77a0b40507f.json

@@ -0,0 +1,600 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_sequential_file_search[openai_client-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_aPe1vS1v5bIwPgl789D5bfmW",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_aPe1vS1v5bIwPgl789D5bfmW",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-690481987689, score: 2.5781234969335522, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-690481987689', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-690481987689|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        },
+        {
+          "role": "assistant",
+          "content": "The Llama 4 Maverick model has 128 experts in its mixture of experts architecture <|file-690481987689|>."
+        },
+        {
+          "role": "user",
+          "content": "Can you tell me more about the architecture?"
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": "call_UqyVZyvMh30eQuiKg7lJIUhQ",
+                    "function": {
+                      "arguments": "",
+                      "name": "knowledge_search"
+                    },
+                    "type": "function"
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "L"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "{\"",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "query",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "rOZPdKYIRKVpUK"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "\":\"",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "KbnFcNwgyiUhKq"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "L",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Wj"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "lama",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "2HMMQqdTQMzWCVE"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " ",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "eF"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "4",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "1C"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " Maver",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "BDs9l6hlr8pF5"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "ick",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " model",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "k0Bkf9cbzy8r0"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " architecture",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "sxpQee"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "\"}",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-127a97b42f23",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "tool_calls",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "j"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/1dbd102b81c7dee87b189664823f2e43f6105ae3111ad9cf843899f495f8bf76.json

@@ -0,0 +1,779 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_compound_or[openai_client-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Show me marketing and sales documents"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_Fvpcub7CINR4M4mne3NrqPSh",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\": \"marketing documents examples\"}"
+              }
+            },
+            {
+              "index": 1,
+              "id": "call_nNONu48z4o4psZIlHNDbWIe6",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\": \"sales documents examples\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_Fvpcub7CINR4M4mne3NrqPSh",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 3 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-862946202236, score: 0.7656461893973538, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-862946202236', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-862946202236|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-862946202234, score: 0.729409669664322, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-862946202234', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-862946202234|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[3] document_id: file-862946202237, score: 0.5528548752393603, attributes: {'region': 'asia', 'category': 'sales', 'date': 1688169600.0, 'filename': 'asia_sales_q3.txt', 'document_id': 'file-862946202237', 'token_count': 17.0, 'metadata_token_count': 31.0} (cite as <|file-862946202237|>)\nAsia Pacific revenue figures for Q3 2023. Record breaking quarter in Asia.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"marketing documents examples\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_nNONu48z4o4psZIlHNDbWIe6",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 3 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-862946202234, score: 0.6624775971970099, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-862946202234', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-862946202234|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-862946202236, score: 0.6241908355663578, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-862946202236', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-862946202236|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "[3] document_id: file-862946202237, score: 0.5761999414500201, attributes: {'region': 'asia', 'category': 'sales', 'date': 1688169600.0, 'filename': 'asia_sales_q3.txt', 'document_id': 'file-862946202237', 'token_count': 17.0, 'metadata_token_count': 31.0} (cite as <|file-862946202237|>)\nAsia Pacific revenue figures for Q3 2023. Record breaking quarter in Asia.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"sales documents examples\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "R6uA5YGaVXy"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "Here are some examples of marketing and sales documents:\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "fvEM0zzsT"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "### Marketing Documents:\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "fM8pbIH2Du"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ywGbbKNyMkSs"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " **European Advertising Campaign**:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "9Df30NePsv"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " This document reports on the results of European advertising campaigns for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "1KDOAYrg"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Ur6YZUx8o2"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " highlighting strong growth in EU markets <|file-862946202236|>.\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "2.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "bGNEay3i7CbT"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " **US Promotional Campaigns**:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "BcVg7O8xUL"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " It details the marketing activities in the US for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "lJMGEMwQQW"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "fWLN9vPA4l"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " showing a ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "hcpod"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% revenue increase in the region <|file-862946202234|>",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "zSOh8pStGzP"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": ".\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "4FFmHXhr"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "### Sales Documents:\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "f2CxS7gaPY"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "hKaxDGxBdyNL"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " **Asia Pacific Revenue Figures**:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "MBZgU7ywir"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " This document provides sales figures for the Asia Pacific region for Q3 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "6qKqjjyo"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "KRSPEkdDeP"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": " noting a record-breaking quarter in Asia <|file-862946202237|>",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "wCkeab"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": ".\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "0FVU0V3j"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": "These documents provide insights into regional marketing strategies and sales performance over specific periods.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "F5QDS9Hw"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-1dbd102b81c7",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "rOg1C0x"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/2443581bf4a71f7b26e3aef9715a65a1cb5ab62249e691e77dcfec2e00be302b.json

@@ -0,0 +1,192 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_compound_and[client_with_models-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What are the engineering updates from the US?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_toDXiP4WpUNQvvvLkTGEbQAr",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"latest engineering updates US 2023\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_toDXiP4WpUNQvvvLkTGEbQAr",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-710505118847, score: 1.4208534597848683, attributes: {'region': 'us', 'category': 'engineering', 'date': 1680307200.0, 'filename': 'us_engineering_q2.txt', 'document_id': 'file-710505118847', 'token_count': 18.0, 'metadata_token_count': 32.0} (cite as <|file-710505118847|>)\nUS technical updates for Q2 2023. New features deployed in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"latest engineering updates US 2023\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2443581bf4a7",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "WaPYOiGjAoJ"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2443581bf4a7",
+          "choices": [
+            {
+              "delta": {
+                "content": "The latest engineering updates from the US include new features deployed in the region for Q2 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "wV3WlUqwS3"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2443581bf4a7",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023 <|file-710505118847|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "LACGsaGI88"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2443581bf4a7",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Rb1SZZK"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/2f3f766a9601f4ffcde5522361da752354dbf42d982c0a98f4f1113f9cc03769.json

@@ -0,0 +1,219 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_non_streaming_file_search[client_with_models-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536-llama_experts]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_90pCu8l9ITbz463ZJxhGGKm3",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_90pCu8l9ITbz463ZJxhGGKm3",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-36441599755, score: 2.933222791810999, attributes: {'filename': 'test_response_non_streaming_file_search.txt', 'document_id': 'file-36441599755', 'token_count': 10.0, 'metadata_token_count': 13.0} (cite as <|file-36441599755|>)\nLlama 4 Maverick has 128 experts\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2f3f766a9601",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ZcjznG6Yo8S"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2f3f766a9601",
+          "choices": [
+            {
+              "delta": {
+                "content": "The Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "otwbPJWhEZ"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2f3f766a9601",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model has ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "WOGpoDlX3rN1"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2f3f766a9601",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts <|file-36441599755|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "td4Hm9RbPE"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-2f3f766a9601",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "PMGTSWK"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/3d3c43d9096b687183989d7ed41915862a01d180e8a90969599d591fec1b8489.json

@@ -0,0 +1,107 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_text_format[client_with_models-txt=openai/gpt-4o-text_format0]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What is the capital of France?"
+        }
+      ],
+      "stream": true
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-3d3c43d9096b",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_eb3c3cb84d",
+          "usage": null,
+          "obfuscation": "F1K7h1lKsaT"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-3d3c43d9096b",
+          "choices": [
+            {
+              "delta": {
+                "content": "The capital of France is Paris.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_eb3c3cb84d",
+          "usage": null,
+          "obfuscation": "Wk7AnksnRP"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-3d3c43d9096b",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_eb3c3cb84d",
+          "usage": null,
+          "obfuscation": "SyYnKNH"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/4eb6e076a5b400260dc58a211bf90a9bd9bb6c940c4d88aba912ae87a8f66816.json

@@ -0,0 +1,250 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_non_streaming_file_search[client_with_models-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536-llama_experts_pdf]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_Zb3UPWm2DOVVTce3d5Uo21FX",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_Zb3UPWm2DOVVTce3d5Uo21FX",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-433408948870, score: 1.4947232325305748, attributes: {'filename': 'llama_stack_and_models.pdf', 'document_id': 'file-433408948870', 'token_count': 98.0, 'metadata_token_count': 11.0} (cite as <|file-433408948870|>)\n,  \nhardware\n \nvendors,\n \nand\n \nAI-focused\n \ncompanies)\n \nthat\n \noffer\n \ntailored\n \ninfrastructure,\n \nsoftware,\n \nand\n \nservices\n \nfor\n \ndeploying\n \nLlama\n \nmodels.\n  \nLlama  4  Maverick  \n Llama  4  Maverick  is  a  Mixture-of-Experts  (MoE)  model  with  17  billion  active  parameters  and  128  experts.   \n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-433408948870, score: 1.1417523389560924, attributes: {'filename': 'llama_stack_and_models.pdf', 'document_id': 'file-433408948870', 'token_count': 498.0, 'metadata_token_count': 11.0} (cite as <|file-433408948870|>)\nLlama  Stack  \nLlama  Stack  Overview  \nLlama  Stack  standardizes  the  core  building  blocks  that  simplify  AI  application  development.  It  codifies  best  \npractices\n \nacross\n \nthe\n \nLlama\n \necosystem.\n \nMore\n \nspecifically,\n \nit\n \nprovides\n  \u25cf  Unified  API  layer  for  Inference,  RAG,  Agents,  Tools,  Safety,  Evals,  and  Telemetry.  \u25cf  Plugin  architecture  to  support  the  rich  ecosystem  of  different  API  implementations  in  various  \nenvironments,\n \nincluding\n \nlocal\n \ndevelopment,\n \non-premises,\n \ncloud,\n \nand\n \nmobile.\n \u25cf  Prepackaged  verified  distributions  which  offer  a  one-stop  solution  for  developers  to  get  started  quickly  \nand\n \nreliably\n \nin\n \nany\n \nenvironment.\n \u25cf  Multiple  developer  interfaces  like  CLI  and  SDKs  for  Python,  Typescript,  iOS,  and  Android.  \u25cf  Standalone  applications  as  examples  for  how  to  build  production-grade  AI  applications  with  Llama  \nStack.\n \nLlama  Stack  Benefits  \n\u25cf  Flexible  Options:  Developers  can  choose  their  preferred  infrastructure  without  changing  APIs  and  enjoy  \nflexible\n \ndeployment\n \nchoices.\n \u25cf  Consistent  Experience:  With  its  unified  APIs,  Llama  Stack  makes  it  easier  to  build,  test,  and  deploy  AI  \napplications\n \nwith\n \nconsistent\n \napplication\n \nbehavior.\n \u25cf  Robust  Ecosystem:  Llama  Stack  is  already  integrated  with  distribution  partners  (cloud  providers,  \nhardware\n \nvendors,\n \nand\n \nAI-focused\n \ncompanies)\n \nthat\n \noffer\n \ntailored\n \ninfrastructure,\n \nsoftware,\n \nand\n \nservices\n \nfor\n \ndeploying\n \nLlama\n \nmodels.\n  \nLlama  4  Maverick  \n Llama  4  Maverick  is  a  Mixture-of-Experts  (MoE)  model  with  17  billion  active  parameters  and  128  experts.   \n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-4eb6e076a5b4",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ot9TaDQyHrX"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-4eb6e076a5b4",
+          "choices": [
+            {
+              "delta": {
+                "content": "The Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "vJfFXqDKPB"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-4eb6e076a5b4",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model is a Mixture-of-Experts (MoE)",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "2LBRj2itsMHM"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-4eb6e076a5b4",
+          "choices": [
+            {
+              "delta": {
+                "content": " model with ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "9Stbjbq"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-4eb6e076a5b4",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts <|file-433408948870|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "KK4ebrfUqW"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-4eb6e076a5b4",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "A1JTNrt"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/53a5ba8ac82376010813f11bb81778b9f05291c964196a545d474c276dcbe405.json

@@ -0,0 +1,202 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_text_format[openai_client-txt=openai/gpt-4o-text_format1]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What is the capital of France?"
+        }
+      ],
+      "response_format": {
+        "type": "json_schema",
+        "json_schema": {
+          "name": "capitals",
+          "schema": {
+            "type": "object",
+            "properties": {
+              "capital": {
+                "type": "string"
+              }
+            }
+          }
+        }
+      },
+      "stream": true
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-53a5ba8ac823",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "zTAPKN8NLoS"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-53a5ba8ac823",
+          "choices": [
+            {
+              "delta": {
+                "content": "{\"",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "3tfleEvigd"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-53a5ba8ac823",
+          "choices": [
+            {
+              "delta": {
+                "content": "capital\":\"",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "FqmKd8"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-53a5ba8ac823",
+          "choices": [
+            {
+              "delta": {
+                "content": "The capital of France is Paris.\"",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "TV7xqUyDie"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-53a5ba8ac823",
+          "choices": [
+            {
+              "delta": {
+                "content": "}",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Ohw3LeZf0TFe"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-53a5ba8ac823",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "8ByAOfX"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/545bde720fe18ee917fd22ff45f434c1e11cfdfec87f51c6d377dd7aaeeb6c5c.json

@@ -0,0 +1,385 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_by_region[client_with_models-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What are the updates from the US region?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_oDB87lp9gQBu3Xr5LRKtgKvc",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"US news updates\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_oDB87lp9gQBu3Xr5LRKtgKvc",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-446066738450, score: 1.0038202356233785, attributes: {'region': 'us', 'category': 'engineering', 'date': 1680307200.0, 'filename': 'us_engineering_q2.txt', 'document_id': 'file-446066738450', 'token_count': 18.0, 'metadata_token_count': 32.0} (cite as <|file-446066738450|>)\nUS technical updates for Q2 2023. New features deployed in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-446066738449, score: 0.7400703615803927, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-446066738449', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-446066738449|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"US news updates\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ww18bkXvZgv"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": "The latest updates from the US region include:\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "rYyrxABNu6"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "8FlLCah0MSt0"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": " Technical updates for Q2 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "phH"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023 with new features deployed in the US region <|file-446066738450|>.\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "jN19v4DXyu"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": "2.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "t74bLfabpn5g"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": " Successful promotional campaigns for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "1D"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ROy4TEmWrk"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": " which resulted in a ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "QKZLoz3"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% increase in revenue <|file-446066738449|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "5rcb3lTUi11"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-545bde720fe1",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "iWXlJa0"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/5d796fd483090eb001e4bbc5e58e0772e062dd86e976ea115e241cf4af29e31c.json

@@ -0,0 +1,725 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_compound_or[client_with_models-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Show me marketing and sales documents"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_XJHVfGOgINI5ssxdPEkIzNGn",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\": \"marketing documents examples\"}"
+              }
+            },
+            {
+              "index": 1,
+              "id": "call_yU3cSA6ZVa05x581uiSYW8De",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\": \"sales documents examples\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_XJHVfGOgINI5ssxdPEkIzNGn",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 3 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-66545595857, score: 0.7657325734983355, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-66545595857', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-66545595857|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-66545595855, score: 0.729409669664322, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-66545595855', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-66545595855|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[3] document_id: file-66545595858, score: 0.5528548752393603, attributes: {'region': 'asia', 'category': 'sales', 'date': 1688169600.0, 'filename': 'asia_sales_q3.txt', 'document_id': 'file-66545595858', 'token_count': 17.0, 'metadata_token_count': 31.0} (cite as <|file-66545595858|>)\nAsia Pacific revenue figures for Q3 2023. Record breaking quarter in Asia.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"marketing documents examples\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_yU3cSA6ZVa05x581uiSYW8De",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 3 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-66545595855, score: 0.6624775971970099, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-66545595855', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-66545595855|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-66545595857, score: 0.624255076486169, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-66545595857', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-66545595857|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "[3] document_id: file-66545595858, score: 0.5761999414500201, attributes: {'region': 'asia', 'category': 'sales', 'date': 1688169600.0, 'filename': 'asia_sales_q3.txt', 'document_id': 'file-66545595858', 'token_count': 17.0, 'metadata_token_count': 31.0} (cite as <|file-66545595858|>)\nAsia Pacific revenue figures for Q3 2023. Record breaking quarter in Asia.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"sales documents examples\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "sCS0WEObq9w"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "Here are some examples of marketing and sales documents:\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "sylalKfmX"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "**Marketing Documents:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "8ZEgciTllZd"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "**\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "dGMJfxh"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "tsPacaJ4eDPD"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": " **European Marketing Document** - This document reports on the European advertising campaign results for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "UZuul9FFnh"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "lXJE322Ble"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": " highlighting strong growth in EU markets <|file-66545595857|>",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": ".\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Dy4yeYTT"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "2.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Yc1ZRuoEhLaD"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": " **US Marketing Document** - This document details US promotional campaigns for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "wL0a8WrT8i"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "g8dTqUQ6iz"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": " noting a ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "GnSI11"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% increase in revenue in the US region <|file-66545595855|>",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "XoC0MGmkMRD"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": ".\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ipOgSJrl"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "**Sales Documents:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ONNWxX50IkZ"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "**\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "1x4JLxe"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "6MVOndt1oixv"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": " **Asia-Pacific Sales Document** - This document provides revenue figures for the Asia Pacific region for Q3 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "NOhDOd2O2e"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Eog46AZ0tJ"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": " marking a record-breaking quarter in Asia <|file-66545595858|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "RTYDY"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-5d796fd48309",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "GETqYY9"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/813ac454f8df150f6a9918ddd8a68279f2e21c8948779350810cdf51987ed559.json

@@ -0,0 +1,219 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_sequential_file_search[client_with_models-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_zS2WxgXWetjnlPt2MzH9Asrc",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_zS2WxgXWetjnlPt2MzH9Asrc",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-5217982280, score: 2.57802841833685, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-5217982280', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-5217982280|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-813ac454f8df",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "kAlKUdypard"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-813ac454f8df",
+          "choices": [
+            {
+              "delta": {
+                "content": "The Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "U9LsCIESSE"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-813ac454f8df",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model has ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "wwn6wHEfw0FY"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-813ac454f8df",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts in its mixture of experts architecture <|file-5217982280|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "k3KlnqkWxH"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-813ac454f8df",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "m4ogAxG"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/89162a717e49e05ac08013f01ad908e76817c51903b5c476b2109f7763820499.json

@@ -54,7 +54,7 @@
                 "tool_calls": [
                   {
                     "index": 0,
-                    "id": "call_qzY7B7EArJwpMqLVer8kcAey",
+                    "id": "call_zS2WxgXWetjnlPt2MzH9Asrc",
                     "function": {
                       "arguments": "",
                       "name": "knowledge_search"
@@ -74,7 +74,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "D"
+          "obfuscation": "u"
         }
       },
       {
@@ -148,7 +148,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "CJiqORTeQGxjIe"
+          "obfuscation": "RFdKkejzJ9k94o"
         }
       },
       {
@@ -185,7 +185,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "r6QBYD3kczuPXu"
+          "obfuscation": "Hb81GJqWa4o3xK"
         }
       },
       {
@@ -222,7 +222,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "QV"
+          "obfuscation": "1F"
         }
       },
       {
@@ -259,7 +259,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "pvqVubGjwneskfA"
+          "obfuscation": "5EIhJTfkplhcF8b"
         }
       },
       {
@@ -296,7 +296,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "xn"
+          "obfuscation": "hy"
         }
       },
       {
@@ -333,7 +333,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "kd"
+          "obfuscation": "Bh"
         }
       },
       {
@@ -370,7 +370,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "sNtz6TsVqvTPw"
+          "obfuscation": "qSHaXAuZaOsnR"
         }
       },
       {
@@ -444,7 +444,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "fBQyzFK1sVnTB"
+          "obfuscation": "oeMApPOcLgemX"
         }
       },
       {
@@ -481,7 +481,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "0U81JnFrkqiB"
+          "obfuscation": "iT8zweOYBMgE"
         }
       },
       {
@@ -555,7 +555,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "s9gav9QCV2Y"
+          "obfuscation": "4QR8kd8nG1V"
         }
       },
       {
@@ -619,7 +619,7 @@
           "service_tier": "default",
           "system_fingerprint": "fp_f64f290af2",
           "usage": null,
-          "obfuscation": "g"
+          "obfuscation": "S"
         }
       }
     ],

tests/integration/responses/recordings/8945025e663d427c80ddf8b982da997647b346b79db66648a1db09214a7ca372.json

@@ -0,0 +1,507 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_by_date_range[client_with_models-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What happened in Q1 2023?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_a9Kxt1z020OLhioOdi8pVAl6",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"events in Q1 2023\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_a9Kxt1z020OLhioOdi8pVAl6",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-664068282482, score: 1.0007501503163803, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-664068282482', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-664068282482|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-664068282484, score: 0.9776032276748411, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-664068282484', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-664068282484|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"events in Q1 2023\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_CwdXa6wdMO3Jbh0oB3NNlIjE",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"key global events Q1 2023\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_CwdXa6wdMO3Jbh0oB3NNlIjE",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-664068282484, score: 0.9341816085356718, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-664068282484', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-664068282484|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-664068282482, score: 0.9132392014678701, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-664068282482', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-664068282482|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"key global events Q1 2023\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "e4Awy8Uo6Cv"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": "In Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "zODOR5zYmdo"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "yQvVB91rkr"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " marketing campaigns in the US and Europe showed positive results.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Kug"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " In the US,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "KgmGx1ov56"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " promotional campaigns led to a revenue increase of ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "V"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% <|file-664068282482|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "jMPOQcwcREa"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " In Europe,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "WSomtA0mMU"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " advertising campaigns resulted in strong growth in EU markets <|file-664068282484|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Z"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " However,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "9Y6mq"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " more details on political,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Bzwxk0pg"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " economic,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "XOGl"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": " or other significant global events weren't found in the current search results.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "M1ALADgkgW"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8945025e663d",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "NDfVBAJ"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/8d448edc2c0023ba3c3498a7e2f03a05b3c38f21e67d49d09aad5935cff13037.json

@@ -0,0 +1,520 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_by_category[client_with_models-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Show me all marketing reports"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_sfoVogRyIflBvHjg8G9ZZ52v",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"marketing reports\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_sfoVogRyIflBvHjg8G9ZZ52v",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-630021438843, score: 0.7788970955409423, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-630021438843', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-630021438843|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-630021438845, score: 0.74726414959878, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-630021438845', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-630021438845|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"marketing reports\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "dxDNR1iXMJd"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "Here are the marketing reports I found:\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "80wR3f26X"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "6KXPlP3ySakz"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": " **US Marketing Report for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "e83B5CRJLU"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023**:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "9FeKTd1pUV"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": " This report highlights that promotional campaigns in the US led to a ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "orXPwx1c"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% increase in revenue for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "SchCp9zvapp"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023 <|file-630021438843|>",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Sccnv1BaF3"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": ".\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Mu1UC4rL"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "2.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "1uBu918De7Ba"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": " **European Marketing Report for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "hH2y5VHDyZ"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023**:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "AXvFJtxuCD"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": " This report discusses the results of advertising campaigns in Europe,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "PedxYuNe"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": " noting strong growth in the EU markets during Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "pfJyqd"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023 <|file-630021438845|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "3UbqhbmFpR"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-8d448edc2c00",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "lYDLApf"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/abb544a926525d5aa863c006abe6dada8aba13f3598b4ddafffc2c861c019c40.json

@@ -0,0 +1,304 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_by_date_range[openai_client-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What happened in Q1 2023?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_PfY7bcXP30PsqQxRzUa1peOc",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Q1 2023 events\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_PfY7bcXP30PsqQxRzUa1peOc",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-151637398235, score: 1.028104385476394, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-151637398235', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-151637398235|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-151637398237, score: 0.9312513654699082, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-151637398237', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-151637398237|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Q1 2023 events\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Bqtevt5MPSd"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": "In Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "dGD9ZZwkokA"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "8EOofwL33Y"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": " promotional campaigns in the United States resulted in a ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "s"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% increase in revenue <|file-151637398235|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "0YA2WkTFqQv"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": " Additionally,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": " there was significant growth in the European markets due to successful advertising campaigns <|file-151637398237|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "cEKaL4U"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-abb544a92652",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "BqZbBfd"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/b615a3e10275f75dfb8da3474c2cbca56862f4a5d67c1b3887c45a80c0fd6bae.json

@@ -0,0 +1,493 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_by_category[openai_client-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Show me all marketing reports"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_54ZqtMaB74gGLhQoT5Kxe1KN",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"marketing reports\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_54ZqtMaB74gGLhQoT5Kxe1KN",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-824569833174, score: 0.7788970955409423, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-824569833174', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-824569833174|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-824569833176, score: 0.74726414959878, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-824569833176', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-824569833176|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"marketing reports\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "S0wk0b3F0v8"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": "Here are two marketing reports:\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "0rdc2O0d6"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Mweu9w6gSOii"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": " **US Marketing Report**:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Q0qYYvVaIN"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": " This report covers promotional campaigns for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "uIhtL8H7"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023 in the US region,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "94u0V02PhV"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": " highlighting a revenue increase of ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% <|file-824569833174|>",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "XnUSenFGx0l"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": ".\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "9utCu2QR"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": "2.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "OKaYfCPWTlYx"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": " **EU Marketing Report**:",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "srw7RTZWcQ"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": " This report details European advertising campaign results for Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "VZI08kBN"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Mwt9DGHXI1"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": " showing strong growth in EU markets <|file-824569833176|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "oYERz"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b615a3e10275",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ICphsuA"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/b6484b82fd6d6cd548e7b31637994cf2bf1d630699b33ff6e5161b880156e840.json

@@ -0,0 +1,291 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_sequential_file_search[client_with_models-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_zS2WxgXWetjnlPt2MzH9Asrc",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_zS2WxgXWetjnlPt2MzH9Asrc",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-5217982280, score: 2.57802841833685, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-5217982280', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-5217982280|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        },
+        {
+          "role": "assistant",
+          "content": "The Llama 4 Maverick model has 128 experts in its mixture of experts architecture <|file-5217982280|>."
+        },
+        {
+          "role": "user",
+          "content": "Can you tell me more about the architecture?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_eC40X8CzoIfT4BUnCbmNGlbT",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model architecture\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_eC40X8CzoIfT4BUnCbmNGlbT",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-5217982280, score: 1.9325686555844606, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-5217982280', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-5217982280|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model architecture\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b6484b82fd6d",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "YG2gnXGln3w"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b6484b82fd6d",
+          "choices": [
+            {
+              "delta": {
+                "content": "The available information only mentions that the Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "zsQxdAF7Bf"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b6484b82fd6d",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model has a mixture of experts architecture with ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "RS3quzkVDo6p"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b6484b82fd6d",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts <|file-5217982280|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "AA2MyYWwqU"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b6484b82fd6d",
+          "choices": [
+            {
+              "delta": {
+                "content": " Further details about its architecture were not provided in the current data.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "PKOKN"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-b6484b82fd6d",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "bZqZWmb"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/c060e0e12e3bf077546d9eec6d4e9d2e080e661291ca9b52e2e4d2721811efc8.json

@@ -0,0 +1,354 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_compound_and[openai_client-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What are the engineering updates from the US?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_xmBCIHb34NLfvnRnt1wDXjie",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"engineering updates US October 2023\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_xmBCIHb34NLfvnRnt1wDXjie",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-90357487492, score: 1.708989789892816, attributes: {'region': 'us', 'category': 'engineering', 'date': 1680307200.0, 'filename': 'us_engineering_q2.txt', 'document_id': 'file-90357487492', 'token_count': 18.0, 'metadata_token_count': 32.0} (cite as <|file-90357487492|>)\nUS technical updates for Q2 2023. New features deployed in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"engineering updates US October 2023\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "QL1wD7Bp6RT"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": "The search for specific engineering updates from the US for October ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "fzZkmiEW1g"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023 did not return the desired results.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "cJGY5dUAi5"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": " It seems that there are only technical updates available for earlier periods,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "KzI4vF1Wib"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": " such as Q2 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "4vlbZJ3Y"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "fdLgA1uk9A"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": " where new features were deployed in the US region <|file-90357487492|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "1NITGWv"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": " For the most current information,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ElPjxSzYB"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": " checking engineering news websites or official announcements is recommended.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "SpqI"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c060e0e12e3b",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "RT09BI6"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/c4915bdee5a93d3c0426367bd29868d84fb3e4b26c372407f177a4812bd90ea7.json

@@ -0,0 +1,250 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_non_streaming_file_search[openai_client-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536-llama_experts_pdf]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_zKJ7WgvCO3tx6yB0We1qI0d8",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_zKJ7WgvCO3tx6yB0We1qI0d8",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-728796232029, score: 1.4945131220963286, attributes: {'filename': 'llama_stack_and_models.pdf', 'document_id': 'file-728796232029', 'token_count': 98.0, 'metadata_token_count': 11.0} (cite as <|file-728796232029|>)\n,  \nhardware\n \nvendors,\n \nand\n \nAI-focused\n \ncompanies)\n \nthat\n \noffer\n \ntailored\n \ninfrastructure,\n \nsoftware,\n \nand\n \nservices\n \nfor\n \ndeploying\n \nLlama\n \nmodels.\n  \nLlama  4  Maverick  \n Llama  4  Maverick  is  a  Mixture-of-Experts  (MoE)  model  with  17  billion  active  parameters  and  128  experts.   \n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-728796232029, score: 1.1415676746925796, attributes: {'filename': 'llama_stack_and_models.pdf', 'document_id': 'file-728796232029', 'token_count': 498.0, 'metadata_token_count': 11.0} (cite as <|file-728796232029|>)\nLlama  Stack  \nLlama  Stack  Overview  \nLlama  Stack  standardizes  the  core  building  blocks  that  simplify  AI  application  development.  It  codifies  best  \npractices\n \nacross\n \nthe\n \nLlama\n \necosystem.\n \nMore\n \nspecifically,\n \nit\n \nprovides\n  \u25cf  Unified  API  layer  for  Inference,  RAG,  Agents,  Tools,  Safety,  Evals,  and  Telemetry.  \u25cf  Plugin  architecture  to  support  the  rich  ecosystem  of  different  API  implementations  in  various  \nenvironments,\n \nincluding\n \nlocal\n \ndevelopment,\n \non-premises,\n \ncloud,\n \nand\n \nmobile.\n \u25cf  Prepackaged  verified  distributions  which  offer  a  one-stop  solution  for  developers  to  get  started  quickly  \nand\n \nreliably\n \nin\n \nany\n \nenvironment.\n \u25cf  Multiple  developer  interfaces  like  CLI  and  SDKs  for  Python,  Typescript,  iOS,  and  Android.  \u25cf  Standalone  applications  as  examples  for  how  to  build  production-grade  AI  applications  with  Llama  \nStack.\n \nLlama  Stack  Benefits  \n\u25cf  Flexible  Options:  Developers  can  choose  their  preferred  infrastructure  without  changing  APIs  and  enjoy  \nflexible\n \ndeployment\n \nchoices.\n \u25cf  Consistent  Experience:  With  its  unified  APIs,  Llama  Stack  makes  it  easier  to  build,  test,  and  deploy  AI  \napplications\n \nwith\n \nconsistent\n \napplication\n \nbehavior.\n \u25cf  Robust  Ecosystem:  Llama  Stack  is  already  integrated  with  distribution  partners  (cloud  providers,  \nhardware\n \nvendors,\n \nand\n \nAI-focused\n \ncompanies)\n \nthat\n \noffer\n \ntailored\n \ninfrastructure,\n \nsoftware,\n \nand\n \nservices\n \nfor\n \ndeploying\n \nLlama\n \nmodels.\n  \nLlama  4  Maverick  \n Llama  4  Maverick  is  a  Mixture-of-Experts  (MoE)  model  with  17  billion  active  parameters  and  128  experts.   \n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c4915bdee5a9",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "P4gkicgUKax"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c4915bdee5a9",
+          "choices": [
+            {
+              "delta": {
+                "content": "The Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "PIPk8kAyow"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c4915bdee5a9",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model is a Mixture-of-Experts (MoE)",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "viELmOBhSxTU"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c4915bdee5a9",
+          "choices": [
+            {
+              "delta": {
+                "content": " model with ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "M866QoC"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c4915bdee5a9",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts <|file-728796232029|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "s6h8QUwLvD"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-c4915bdee5a9",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ibcjGHS"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/d07dea1671082e52174ac8d467f3e20958091c2ccc81c976e0a3f26b3f27c27e.json

@@ -0,0 +1,600 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_sequential_file_search[client_with_models-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_zS2WxgXWetjnlPt2MzH9Asrc",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_zS2WxgXWetjnlPt2MzH9Asrc",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-5217982280, score: 2.57802841833685, attributes: {'filename': 'test_sequential_file_search.txt', 'document_id': 'file-5217982280', 'token_count': 19.0, 'metadata_token_count': 11.0} (cite as <|file-5217982280|>)\nThe Llama 4 Maverick model has 128 experts in its mixture of experts architecture.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        },
+        {
+          "role": "assistant",
+          "content": "The Llama 4 Maverick model has 128 experts in its mixture of experts architecture <|file-5217982280|>."
+        },
+        {
+          "role": "user",
+          "content": "Can you tell me more about the architecture?"
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": "call_eC40X8CzoIfT4BUnCbmNGlbT",
+                    "function": {
+                      "arguments": "",
+                      "name": "knowledge_search"
+                    },
+                    "type": "function"
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "1"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "{\"",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "query",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "WaHwUohOgfnJOH"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "\":\"",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "OWZ9cPJiOvgHp6"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "L",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "0c"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "lama",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "rjJXIDI6BKl39ZI"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " ",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "iM"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "4",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "15"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " Maver",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "bsgmAqW9qyRkB"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "ick",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " model",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "0y5WwPvUi2J2l"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " architecture",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "8YsQud"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "\"}",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d07dea167108",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "tool_calls",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "z"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/d1c19e856c523b3f52754ce4c79690a12c451726e76c7f51d623804405234402.json

@@ -0,0 +1,107 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_text_format[openai_client-txt=openai/gpt-4o-text_format0]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What is the capital of France?"
+        }
+      ],
+      "stream": true
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d1c19e856c52",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "GmDyFBTczH2"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d1c19e856c52",
+          "choices": [
+            {
+              "delta": {
+                "content": "The capital of France is Paris.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "o24r1KRzNw"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d1c19e856c52",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "igzuBfl"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/d4b919f92aa0cfebf48e2ecd7f3b82014e164e9ec8d06310de8e984e15c704c1.json

@@ -0,0 +1,466 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_by_region[openai_client-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What are the updates from the US region?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_tH5vt4iTTESjlL0vKwDKKLtU",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"latest updates from the US region\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_tH5vt4iTTESjlL0vKwDKKLtU",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-247992711531, score: 1.746917709229671, attributes: {'region': 'us', 'category': 'engineering', 'date': 1680307200.0, 'filename': 'us_engineering_q2.txt', 'document_id': 'file-247992711531', 'token_count': 18.0, 'metadata_token_count': 32.0} (cite as <|file-247992711531|>)\nUS technical updates for Q2 2023. New features deployed in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-247992711530, score: 0.9418284563822891, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-247992711530', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-247992711530|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"latest updates from the US region\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "LVBYUvJmYrP"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": "Here are some recent updates related to the US region:\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "I0WY1gRg2"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": "1.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "AFHfwzJAJYrn"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": " In Q2 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "6Tu1Zkc5Cl"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "FM2xXQk4Xg"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": " new technical features were deployed in the US region,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "JWXl941NV"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": " indicating progress in engineering and technical advancements <|file-247992711531|>",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Cf"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": ".\n\n",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "rTeLdej6"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": "2.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "0CRQo167cVwS"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": " For Q1 ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "io5dVAPR6"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": "2023,",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "JEEfXMv3tZ"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": " there were notable marketing activities with promotional campaigns leading to a ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "jS6I4Aw"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": "15% increase in revenue in the US region <|file-247992711530|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "UFLsAL2aZCu"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-d4b919f92aa0",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "Nk1jxtn"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/de35ec2b1494372e2054e04ac119a47e104212c0540f840632521852071b18ef.json

@@ -0,0 +1,202 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_text_format[client_with_models-txt=openai/gpt-4o-text_format1]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What is the capital of France?"
+        }
+      ],
+      "response_format": {
+        "type": "json_schema",
+        "json_schema": {
+          "name": "capitals",
+          "schema": {
+            "type": "object",
+            "properties": {
+              "capital": {
+                "type": "string"
+              }
+            }
+          }
+        }
+      },
+      "stream": true
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-de35ec2b1494",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "raD8PNyHuff"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-de35ec2b1494",
+          "choices": [
+            {
+              "delta": {
+                "content": "{\"",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "bKNctm8IOW"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-de35ec2b1494",
+          "choices": [
+            {
+              "delta": {
+                "content": "capital\":\"",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "sCtYBe"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-de35ec2b1494",
+          "choices": [
+            {
+              "delta": {
+                "content": "The capital of France is Paris.\"",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "CfKJfvFsG3"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-de35ec2b1494",
+          "choices": [
+            {
+              "delta": {
+                "content": "}",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "5CAMRCsJP8iK"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-de35ec2b1494",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_cbf1785567",
+          "usage": null,
+          "obfuscation": "GqXk1WL"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/e69443fa2e446828bc12b839dacc009739a2feffea573f452a1f0af8e6e4be6e.json

@@ -0,0 +1,596 @@
+{
+  "test_id": "tests/integration/responses/test_file_search.py::test_response_file_search_filter_by_date_range[client_with_models-txt=openai/gpt-4o]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "What happened in Q1 2023?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_a9Kxt1z020OLhioOdi8pVAl6",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"events in Q1 2023\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_a9Kxt1z020OLhioOdi8pVAl6",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 2 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-664068282482, score: 1.0007501503163803, attributes: {'region': 'us', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'us_marketing_q1.txt', 'document_id': 'file-664068282482', 'token_count': 21.0, 'metadata_token_count': 32.0} (cite as <|file-664068282482|>)\nUS promotional campaigns for Q1 2023. Revenue increased by 15% in the US region.\n"
+            },
+            {
+              "type": "text",
+              "text": "[2] document_id: file-664068282484, score: 0.9776032276748411, attributes: {'region': 'eu', 'category': 'marketing', 'date': 1672531200.0, 'filename': 'eu_marketing_q1.txt', 'document_id': 'file-664068282484', 'token_count': 17.0, 'metadata_token_count': 32.0} (cite as <|file-664068282484|>)\nEuropean advertising campaign results for Q1 2023. Strong growth in EU markets.\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"events in Q1 2023\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": "call_CwdXa6wdMO3Jbh0oB3NNlIjE",
+                    "function": {
+                      "arguments": "",
+                      "name": "knowledge_search"
+                    },
+                    "type": "function"
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "r"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "{\"",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "query",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "sJPYFmJf6VNnKV"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "\":\"",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "hPZ5ojU9O1JCwk"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "key",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " global",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "PnDPR1Y8zCab"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " events",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "ePyNFvW19GBw"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " Q",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "H"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "1",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "4D"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": " ",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "IA"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "202",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "3",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "qR"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": [
+                  {
+                    "index": 0,
+                    "id": null,
+                    "function": {
+                      "arguments": "\"}",
+                      "name": null
+                    },
+                    "type": null
+                  }
+                ]
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": ""
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-e69443fa2e44",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "tool_calls",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "n"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}

tests/integration/responses/recordings/fbc3663df932c59c014cfcc0673ff0747040227532b0896782a36c10aa78803a.json

@@ -0,0 +1,219 @@
+{
+  "test_id": "tests/integration/responses/test_tool_responses.py::test_response_non_streaming_file_search[openai_client-txt=openai/gpt-4o:emb=openai/text-embedding-3-small:dim=1536-llama_experts]",
+  "request": {
+    "method": "POST",
+    "url": "https://api.openai.com/v1/v1/chat/completions",
+    "headers": {},
+    "body": {
+      "model": "gpt-4o",
+      "messages": [
+        {
+          "role": "user",
+          "content": "How many experts does the Llama 4 Maverick model have?"
+        },
+        {
+          "role": "assistant",
+          "content": "",
+          "tool_calls": [
+            {
+              "index": 0,
+              "id": "call_9Ofp0Uepi3uOnEmuFtm9yvCU",
+              "type": "function",
+              "function": {
+                "name": "knowledge_search",
+                "arguments": "{\"query\":\"Llama 4 Maverick model number of experts\"}"
+              }
+            }
+          ]
+        },
+        {
+          "role": "tool",
+          "tool_call_id": "call_9Ofp0Uepi3uOnEmuFtm9yvCU",
+          "content": [
+            {
+              "type": "text",
+              "text": "knowledge_search tool found 1 chunks:\nBEGIN of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "[1] document_id: file-78420035045, score: 2.933222791810999, attributes: {'filename': 'test_response_non_streaming_file_search.txt', 'document_id': 'file-78420035045', 'token_count': 10.0, 'metadata_token_count': 13.0} (cite as <|file-78420035045|>)\nLlama 4 Maverick has 128 experts\n"
+            },
+            {
+              "type": "text",
+              "text": "END of knowledge_search tool results.\n"
+            },
+            {
+              "type": "text",
+              "text": "The above results were retrieved to help answer the user's query: \"Llama 4 Maverick model number of experts\". Use them as supporting information only in answering this query. Cite sources immediately at the end of sentences before punctuation, using `<|file-id|>` format (e.g., 'This is a fact <|file-Cn3MSNn72ENTiiq11Qda4A|>.'). Do not add extra punctuation. Use only the file IDs provided (do not invent new ones).\n"
+            }
+          ]
+        }
+      ],
+      "stream": true,
+      "tools": [
+        {
+          "type": "function",
+          "function": {
+            "name": "knowledge_search",
+            "description": "Search for information in a database.",
+            "parameters": {
+              "type": "object",
+              "properties": {
+                "query": {
+                  "type": "string",
+                  "description": "The query to search for. Can be a natural language sentence or keywords."
+                }
+              },
+              "required": [
+                "query"
+              ]
+            }
+          }
+        }
+      ]
+    },
+    "endpoint": "/v1/chat/completions",
+    "model": "gpt-4o"
+  },
+  "response": {
+    "body": [
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-fbc3663df932",
+          "choices": [
+            {
+              "delta": {
+                "content": "",
+                "function_call": null,
+                "refusal": null,
+                "role": "assistant",
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "jPCn755gSJm"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-fbc3663df932",
+          "choices": [
+            {
+              "delta": {
+                "content": "The Llama ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "7DiMbaNjQh"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-fbc3663df932",
+          "choices": [
+            {
+              "delta": {
+                "content": "4 Maverick model has ",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "YHm5Z1DAbROO"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-fbc3663df932",
+          "choices": [
+            {
+              "delta": {
+                "content": "128 experts <|file-78420035045|>.",
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": null,
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "BC68VJcdY6"
+        }
+      },
+      {
+        "__type__": "openai.types.chat.chat_completion_chunk.ChatCompletionChunk",
+        "__data__": {
+          "id": "rec-fbc3663df932",
+          "choices": [
+            {
+              "delta": {
+                "content": null,
+                "function_call": null,
+                "refusal": null,
+                "role": null,
+                "tool_calls": null
+              },
+              "finish_reason": "stop",
+              "index": 0,
+              "logprobs": null
+            }
+          ],
+          "created": 0,
+          "model": "gpt-4o-2024-08-06",
+          "object": "chat.completion.chunk",
+          "service_tier": "default",
+          "system_fingerprint": "fp_f64f290af2",
+          "usage": null,
+          "obfuscation": "2y0NJct"
+        }
+      }
+    ],
+    "is_streaming": true
+  },
+  "id_normalization_mapping": {}
+}