(Feat) - Show Error Logs on LiteLLM UI (#8904)

* fix test_moderations_bad_model

* use async_post_call_failure_hook

* basic logging errors in DB

* show status on ui

* show status on ui

* ui show request / response side by side

* stash fixes

* working, track raw request

* track error info in metadata

* fix showing error / request / response logs

* show traceback on error viewer

* ui with traceback of error

* fix async_post_call_failure_hook

* fix(http_parsing_utils.py): orjson can throw errors on some emoji's in text, default to json.loads

* test_get_error_information

* fix code quality

* rename proxy track cost callback test

* _should_store_errors_in_spend_logs

* feature flag error logs

* Revert "_should_store_errors_in_spend_logs"

This reverts commit 7f345df477.

* Revert "feature flag error logs"

This reverts commit 0e90c022bb.

* test_spend_logs_payload

* fix OTEL log_db_metrics

* fix import json

* fix ui linting error

* test_async_post_call_failure_hook

* test_chat_completion_bad_model_with_spend_logs

---------

Co-authored-by: Krrish Dholakia <krrishdholakia@gmail.com>

docs/my-website/docs/proxy/architecture.md

@@ -36,7 +36,7 @@ import TabItem from '@theme/TabItem';
         - Virtual Key Rate Limit
         - User Rate Limit
         - Team Limit
-    - The `_PROXY_track_cost_callback` updates spend / usage in the LiteLLM database. [Here is everything tracked in the DB per request](https://github.com/BerriAI/litellm/blob/ba41a72f92a9abf1d659a87ec880e8e319f87481/schema.prisma#L172)
+    - The `_ProxyDBLogger` updates spend / usage in the LiteLLM database. [Here is everything tracked in the DB per request](https://github.com/BerriAI/litellm/blob/ba41a72f92a9abf1d659a87ec880e8e319f87481/schema.prisma#L172)
 
 ## Frequently Asked Questions
 

litellm/litellm_core_utils/litellm_logging.py

@@ -3114,10 +3114,26 @@ class StandardLoggingPayloadSetup:
             str(original_exception.__class__.__name__) if original_exception else ""
         )
         _llm_provider_in_exception = getattr(original_exception, "llm_provider", "")
+
+        # Get traceback information (first 100 lines)
+        traceback_info = ""
+        if original_exception:
+            tb = getattr(original_exception, "__traceback__", None)
+            if tb:
+                import traceback
+
+                tb_lines = traceback.format_tb(tb)
+                traceback_info = "".join(tb_lines[:100])  # Limit to first 100 lines
+
+        # Get additional error details
+        error_message = str(original_exception)
+
         return StandardLoggingPayloadErrorInformation(
             error_code=error_status,
             error_class=error_class,
             llm_provider=_llm_provider_in_exception,
+            traceback=traceback_info,
+            error_message=error_message if original_exception else "",
         )
 
     @staticmethod

litellm/proxy/_types.py

@@ -26,6 +26,8 @@ from litellm.types.utils import (
     ModelResponse,
     ProviderField,
     StandardCallbackDynamicParams,
+    StandardLoggingPayloadErrorInformation,
+    StandardLoggingPayloadStatus,
     StandardPassThroughResponseObject,
     TextCompletionResponse,
 )
@@ -1854,6 +1856,8 @@ class SpendLogsMetadata(TypedDict):
     ]  # special param to log k,v pairs to spendlogs for a call
     requester_ip_address: Optional[str]
     applied_guardrails: Optional[List[str]]
+    status: StandardLoggingPayloadStatus
+    error_information: Optional[StandardLoggingPayloadErrorInformation]
 
 
 class SpendLogsPayload(TypedDict):

litellm/proxy/db/log_db_metrics.py

@@ -64,10 +64,10 @@ def log_db_metrics(func):
                 # in litellm custom callbacks kwargs is passed as arg[0]
                 # https://docs.litellm.ai/docs/observability/custom_callback#callback-functions
                 args is not None
-                and len(args) > 0
+                and len(args) > 1
-                and isinstance(args[0], dict)
+                and isinstance(args[1], dict)
             ):
-                passed_kwargs = args[0]
+                passed_kwargs = args[1]
                 parent_otel_span = _get_parent_otel_span_from_kwargs(
                     kwargs=passed_kwargs
                 )

litellm/proxy/hooks/proxy_track_cost_callback.py

@@ -1,138 +1,205 @@
 import asyncio
 import traceback
-from typing import Optional, Union, cast
+from datetime import datetime
+from typing import Any, Optional, Union, cast
 
 import litellm
 from litellm._logging import verbose_proxy_logger
+from litellm.integrations.custom_logger import CustomLogger
 from litellm.litellm_core_utils.core_helpers import (
     _get_parent_otel_span_from_kwargs,
     get_litellm_metadata_from_kwargs,
 )
+from litellm.litellm_core_utils.litellm_logging import StandardLoggingPayloadSetup
+from litellm.proxy._types import UserAPIKeyAuth
 from litellm.proxy.auth.auth_checks import log_db_metrics
-from litellm.types.utils import StandardLoggingPayload
+from litellm.types.utils import (
+    StandardLoggingPayload,
+    StandardLoggingUserAPIKeyMetadata,
+)
 from litellm.utils import get_end_user_id_for_cost_tracking
 
 
-@log_db_metrics
+class _ProxyDBLogger(CustomLogger):
-async def _PROXY_track_cost_callback(
+    async def async_log_success_event(self, kwargs, response_obj, start_time, end_time):
-    kwargs,  # kwargs to completion
+        await self._PROXY_track_cost_callback(
-    completion_response: litellm.ModelResponse,  # response from completion
+            kwargs, response_obj, start_time, end_time
-    start_time=None,
-    end_time=None,  # start/end time for completion
-):
-    from litellm.proxy.proxy_server import (
-        prisma_client,
-        proxy_logging_obj,
-        update_cache,
-        update_database,
-    )
-
-    verbose_proxy_logger.debug("INSIDE _PROXY_track_cost_callback")
-    try:
-        verbose_proxy_logger.debug(
-            f"kwargs stream: {kwargs.get('stream', None)} + complete streaming response: {kwargs.get('complete_streaming_response', None)}"
-        )
-        parent_otel_span = _get_parent_otel_span_from_kwargs(kwargs=kwargs)
-        litellm_params = kwargs.get("litellm_params", {}) or {}
-        end_user_id = get_end_user_id_for_cost_tracking(litellm_params)
-        metadata = get_litellm_metadata_from_kwargs(kwargs=kwargs)
-        user_id = cast(Optional[str], metadata.get("user_api_key_user_id", None))
-        team_id = cast(Optional[str], metadata.get("user_api_key_team_id", None))
-        org_id = cast(Optional[str], metadata.get("user_api_key_org_id", None))
-        key_alias = cast(Optional[str], metadata.get("user_api_key_alias", None))
-        end_user_max_budget = metadata.get("user_api_end_user_max_budget", None)
-        sl_object: Optional[StandardLoggingPayload] = kwargs.get(
-            "standard_logging_object", None
-        )
-        response_cost = (
-            sl_object.get("response_cost", None)
-            if sl_object is not None
-            else kwargs.get("response_cost", None)
         )
 
-        if response_cost is not None:
+    async def async_post_call_failure_hook(
-            user_api_key = metadata.get("user_api_key", None)
+        self,
-            if kwargs.get("cache_hit", False) is True:
+        request_data: dict,
-                response_cost = 0.0
+        original_exception: Exception,
-                verbose_proxy_logger.info(
+        user_api_key_dict: UserAPIKeyAuth,
-                    f"Cache Hit: response_cost {response_cost}, for user_id {user_id}"
+    ):
-                )
+        from litellm.proxy.proxy_server import update_database
 
-            verbose_proxy_logger.debug(
+        _metadata = dict(
-                f"user_api_key {user_api_key}, prisma_client: {prisma_client}"
+            StandardLoggingUserAPIKeyMetadata(
+                user_api_key_hash=user_api_key_dict.api_key,
+                user_api_key_alias=user_api_key_dict.key_alias,
+                user_api_key_user_email=user_api_key_dict.user_email,
+                user_api_key_user_id=user_api_key_dict.user_id,
+                user_api_key_team_id=user_api_key_dict.team_id,
+                user_api_key_org_id=user_api_key_dict.org_id,
+                user_api_key_team_alias=user_api_key_dict.team_alias,
+                user_api_key_end_user_id=user_api_key_dict.end_user_id,
             )
-            if _should_track_cost_callback(
+        )
-                user_api_key=user_api_key,
+        _metadata["user_api_key"] = user_api_key_dict.api_key
-                user_id=user_id,
+        _metadata["status"] = "failure"
-                team_id=team_id,
+        _metadata["error_information"] = (
-                end_user_id=end_user_id,
+            StandardLoggingPayloadSetup.get_error_information(
-            ):
+                original_exception=original_exception,
-                ## UPDATE DATABASE
+            )
-                await update_database(
+        )
-                    token=user_api_key,
-                    response_cost=response_cost,
-                    user_id=user_id,
-                    end_user_id=end_user_id,
-                    team_id=team_id,
-                    kwargs=kwargs,
-                    completion_response=completion_response,
-                    start_time=start_time,
-                    end_time=end_time,
-                    org_id=org_id,
-                )
 
-                # update cache
+        existing_metadata: dict = request_data.get("metadata", None) or {}
-                asyncio.create_task(
+        existing_metadata.update(_metadata)
-                    update_cache(
+        existing_metadata["proxy_server_request"] = (
+            request_data.get("proxy_server_request", {}) or {}
+        )
+        request_data["litellm_params"] = {}
+        request_data["litellm_params"]["metadata"] = existing_metadata
+
+        await update_database(
+            token=user_api_key_dict.api_key,
+            response_cost=0.0,
+            user_id=user_api_key_dict.user_id,
+            end_user_id=user_api_key_dict.end_user_id,
+            team_id=user_api_key_dict.team_id,
+            kwargs=request_data,
+            completion_response=original_exception,
+            start_time=datetime.now(),
+            end_time=datetime.now(),
+            org_id=user_api_key_dict.org_id,
+        )
+
+    @log_db_metrics
+    async def _PROXY_track_cost_callback(
+        self,
+        kwargs,  # kwargs to completion
+        completion_response: Optional[
+            Union[litellm.ModelResponse, Any]
+        ],  # response from completion
+        start_time=None,
+        end_time=None,  # start/end time for completion
+    ):
+        from litellm.proxy.proxy_server import (
+            prisma_client,
+            proxy_logging_obj,
+            update_cache,
+            update_database,
+        )
+
+        verbose_proxy_logger.debug("INSIDE _PROXY_track_cost_callback")
+        try:
+            verbose_proxy_logger.debug(
+                f"kwargs stream: {kwargs.get('stream', None)} + complete streaming response: {kwargs.get('complete_streaming_response', None)}"
+            )
+            parent_otel_span = _get_parent_otel_span_from_kwargs(kwargs=kwargs)
+            litellm_params = kwargs.get("litellm_params", {}) or {}
+            end_user_id = get_end_user_id_for_cost_tracking(litellm_params)
+            metadata = get_litellm_metadata_from_kwargs(kwargs=kwargs)
+            user_id = cast(Optional[str], metadata.get("user_api_key_user_id", None))
+            team_id = cast(Optional[str], metadata.get("user_api_key_team_id", None))
+            org_id = cast(Optional[str], metadata.get("user_api_key_org_id", None))
+            key_alias = cast(Optional[str], metadata.get("user_api_key_alias", None))
+            end_user_max_budget = metadata.get("user_api_end_user_max_budget", None)
+            sl_object: Optional[StandardLoggingPayload] = kwargs.get(
+                "standard_logging_object", None
+            )
+            response_cost = (
+                sl_object.get("response_cost", None)
+                if sl_object is not None
+                else kwargs.get("response_cost", None)
+            )
+
+            if response_cost is not None:
+                user_api_key = metadata.get("user_api_key", None)
+                if kwargs.get("cache_hit", False) is True:
+                    response_cost = 0.0
+                    verbose_proxy_logger.info(
+                        f"Cache Hit: response_cost {response_cost}, for user_id {user_id}"
+                    )
+
+                verbose_proxy_logger.debug(
+                    f"user_api_key {user_api_key}, prisma_client: {prisma_client}"
+                )
+                if _should_track_cost_callback(
+                    user_api_key=user_api_key,
+                    user_id=user_id,
+                    team_id=team_id,
+                    end_user_id=end_user_id,
+                ):
+                    ## UPDATE DATABASE
+                    await update_database(
                         token=user_api_key,
+                        response_cost=response_cost,
                         user_id=user_id,
                         end_user_id=end_user_id,
-                        response_cost=response_cost,
                         team_id=team_id,
-                        parent_otel_span=parent_otel_span,
+                        kwargs=kwargs,
+                        completion_response=completion_response,
+                        start_time=start_time,
+                        end_time=end_time,
+                        org_id=org_id,
                     )
-                )
 
-                await proxy_logging_obj.slack_alerting_instance.customer_spend_alert(
+                    # update cache
-                    token=user_api_key,
+                    asyncio.create_task(
-                    key_alias=key_alias,
+                        update_cache(
-                    end_user_id=end_user_id,
+                            token=user_api_key,
-                    response_cost=response_cost,
+                            user_id=user_id,
-                    max_budget=end_user_max_budget,
+                            end_user_id=end_user_id,
-                )
+                            response_cost=response_cost,
-            else:
+                            team_id=team_id,
-                raise Exception(
+                            parent_otel_span=parent_otel_span,
-                    "User API key and team id and user id missing from custom callback."
+                        )
-                )
+                    )
-        else:
+
-            if kwargs["stream"] is not True or (
+                    await proxy_logging_obj.slack_alerting_instance.customer_spend_alert(
-                kwargs["stream"] is True and "complete_streaming_response" in kwargs
+                        token=user_api_key,
-            ):
+                        key_alias=key_alias,
-                if sl_object is not None:
+                        end_user_id=end_user_id,
-                    cost_tracking_failure_debug_info: Union[dict, str] = (
+                        response_cost=response_cost,
-                        sl_object["response_cost_failure_debug_info"]  # type: ignore
+                        max_budget=end_user_max_budget,
-                        or "response_cost_failure_debug_info is None in standard_logging_object"
                     )
                 else:
-                    cost_tracking_failure_debug_info = (
+                    raise Exception(
-                        "standard_logging_object not found"
+                        "User API key and team id and user id missing from custom callback."
                     )
-                model = kwargs.get("model")
+            else:
-                raise Exception(
+                if kwargs["stream"] is not True or (
-                    f"Cost tracking failed for model={model}.\nDebug info - {cost_tracking_failure_debug_info}\nAdd custom pricing - https://docs.litellm.ai/docs/proxy/custom_pricing"
+                    kwargs["stream"] is True and "complete_streaming_response" in kwargs
+                ):
+                    if sl_object is not None:
+                        cost_tracking_failure_debug_info: Union[dict, str] = (
+                            sl_object["response_cost_failure_debug_info"]  # type: ignore
+                            or "response_cost_failure_debug_info is None in standard_logging_object"
+                        )
+                    else:
+                        cost_tracking_failure_debug_info = (
+                            "standard_logging_object not found"
+                        )
+                    model = kwargs.get("model")
+                    raise Exception(
+                        f"Cost tracking failed for model={model}.\nDebug info - {cost_tracking_failure_debug_info}\nAdd custom pricing - https://docs.litellm.ai/docs/proxy/custom_pricing"
+                    )
+        except Exception as e:
+            error_msg = f"Error in tracking cost callback - {str(e)}\n Traceback:{traceback.format_exc()}"
+            model = kwargs.get("model", "")
+            metadata = kwargs.get("litellm_params", {}).get("metadata", {})
+            error_msg += f"\n Args to _PROXY_track_cost_callback\n model: {model}\n metadata: {metadata}\n"
+            asyncio.create_task(
+                proxy_logging_obj.failed_tracking_alert(
+                    error_message=error_msg,
+                    failing_model=model,
                 )
-    except Exception as e:
-        error_msg = f"Error in tracking cost callback - {str(e)}\n Traceback:{traceback.format_exc()}"
-        model = kwargs.get("model", "")
-        metadata = kwargs.get("litellm_params", {}).get("metadata", {})
-        error_msg += f"\n Args to _PROXY_track_cost_callback\n model: {model}\n metadata: {metadata}\n"
-        asyncio.create_task(
-            proxy_logging_obj.failed_tracking_alert(
-                error_message=error_msg,
-                failing_model=model,
             )
-        )
+            verbose_proxy_logger.exception(
-        verbose_proxy_logger.exception("Error in tracking cost callback - %s", str(e))
+                "Error in tracking cost callback - %s", str(e)
+            )
 
 
 def _should_track_cost_callback(

litellm/proxy/proxy_config.yaml

@@ -8,4 +8,5 @@ model_list:
 
 general_settings:
   store_model_in_db: true
+  store_prompts_in_spend_logs: true
 

litellm/proxy/proxy_server.py

@@ -114,6 +114,7 @@ from litellm.litellm_core_utils.core_helpers import (
     _get_parent_otel_span_from_kwargs,
     get_litellm_metadata_from_kwargs,
 )
+from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
 from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler, HTTPHandler
 from litellm.proxy._types import *
 from litellm.proxy.analytics_endpoints.analytics_endpoints import (
@@ -178,7 +179,7 @@ from litellm.proxy.hooks.prompt_injection_detection import (
     _OPTIONAL_PromptInjectionDetection,
 )
 from litellm.proxy.hooks.proxy_failure_handler import _PROXY_failure_handler
-from litellm.proxy.hooks.proxy_track_cost_callback import _PROXY_track_cost_callback
+from litellm.proxy.hooks.proxy_track_cost_callback import _ProxyDBLogger
 from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request
 from litellm.proxy.management_endpoints.budget_management_endpoints import (
     router as budget_management_router,
@@ -937,10 +938,7 @@ def load_from_azure_key_vault(use_azure_key_vault: bool = False):
 def cost_tracking():
     global prisma_client
     if prisma_client is not None:
-        if isinstance(litellm._async_success_callback, list):
+        litellm.logging_callback_manager.add_litellm_callback(_ProxyDBLogger())
-            verbose_proxy_logger.debug("setting litellm success callback to track cost")
-            if (_PROXY_track_cost_callback) not in litellm._async_success_callback:  # type: ignore
-                litellm.logging_callback_manager.add_litellm_async_success_callback(_PROXY_track_cost_callback)  # type: ignore
 
 
 def error_tracking():
@@ -3727,9 +3725,14 @@ async def chat_completion(  # noqa: PLR0915
         timeout = getattr(
             e, "timeout", None
         )  # returns the timeout set by the wrapper. Used for testing if model-specific timeout are set correctly
-
+        _litellm_logging_obj: Optional[LiteLLMLoggingObj] = data.get(
+            "litellm_logging_obj", None
+        )
         custom_headers = get_custom_headers(
             user_api_key_dict=user_api_key_dict,
+            call_id=(
+                _litellm_logging_obj.litellm_call_id if _litellm_logging_obj else None
+            ),
             version=version,
             response_cost=0,
             model_region=getattr(user_api_key_dict, "allowed_model_region", ""),

litellm/proxy/spend_tracking/spend_tracking_utils.py

@@ -47,6 +47,8 @@ def _get_spend_logs_metadata(
             requester_ip_address=None,
             additional_usage_values=None,
             applied_guardrails=None,
+            status=None or "success",
+            error_information=None,
         )
     verbose_proxy_logger.debug(
         "getting payload for SpendLogs, available keys in metadata: "
@@ -161,7 +163,6 @@ def get_logging_payload(  # noqa: PLR0915
         import time
 
         id = f"{id}_cache_hit{time.time()}"  # SpendLogs does not allow duplicate request_id
-
     try:
         payload: SpendLogsPayload = SpendLogsPayload(
             request_id=str(id),
@@ -193,7 +194,9 @@ def get_logging_payload(  # noqa: PLR0915
             model_id=_model_id,
             requester_ip_address=clean_metadata.get("requester_ip_address", None),
             custom_llm_provider=kwargs.get("custom_llm_provider", ""),
-            messages=_get_messages_for_spend_logs_payload(standard_logging_payload),
+            messages=_get_messages_for_spend_logs_payload(
+                standard_logging_payload=standard_logging_payload, metadata=metadata
+            ),
             response=_get_response_for_spend_logs_payload(standard_logging_payload),
         )
 
@@ -293,12 +296,19 @@ async def get_spend_by_team_and_customer(
 
 
 def _get_messages_for_spend_logs_payload(
-    payload: Optional[StandardLoggingPayload],
+    standard_logging_payload: Optional[StandardLoggingPayload],
+    metadata: Optional[dict] = None,
 ) -> str:
-    if payload is None:
-        return "{}"
     if _should_store_prompts_and_responses_in_spend_logs():
-        return json.dumps(payload.get("messages", {}))
+        metadata = metadata or {}
+        if metadata.get("status", None) == "failure":
+            _proxy_server_request = metadata.get("proxy_server_request", {})
+            _request_body = _proxy_server_request.get("body", {}) or {}
+            return json.dumps(_request_body, default=str)
+        else:
+            if standard_logging_payload is None:
+                return "{}"
+            return json.dumps(standard_logging_payload.get("messages", {}))
     return "{}"
 
 

litellm/proxy/utils.py

@@ -784,13 +784,17 @@ class ProxyLogging:
                 else:
                     _callback = callback  # type: ignore
                 if _callback is not None and isinstance(_callback, CustomLogger):
-                    await _callback.async_post_call_failure_hook(
+                    asyncio.create_task(
-                        request_data=request_data,
+                        _callback.async_post_call_failure_hook(
-                        user_api_key_dict=user_api_key_dict,
+                            request_data=request_data,
-                        original_exception=original_exception,
+                            user_api_key_dict=user_api_key_dict,
+                            original_exception=original_exception,
+                        )
                     )
             except Exception as e:
-                raise e
+                verbose_proxy_logger.exception(
+                    f"[Non-Blocking] Error in post_call_failure_hook: {e}"
+                )
         return
 
     def _is_proxy_only_error(

litellm/types/utils.py

@@ -1606,6 +1606,8 @@ class StandardLoggingPayloadErrorInformation(TypedDict, total=False):
     error_code: Optional[str]
     error_class: Optional[str]
     llm_provider: Optional[str]
+    traceback: Optional[str]
+    error_message: Optional[str]
 
 
 class StandardLoggingGuardrailInformation(TypedDict, total=False):

tests/litellm/proxy/hooks/test_proxy_track_cost_callback.py

@@ -0,0 +1,80 @@
+import json
+import os
+import sys
+
+import pytest
+from fastapi.testclient import TestClient
+
+sys.path.insert(
+    0, os.path.abspath("../../../..")
+)  # Adds the parent directory to the system path
+
+from datetime import datetime
+from unittest.mock import AsyncMock, MagicMock, patch
+
+from litellm.proxy._types import UserAPIKeyAuth
+from litellm.proxy.hooks.proxy_track_cost_callback import _ProxyDBLogger
+from litellm.types.utils import StandardLoggingPayload
+
+
+@pytest.mark.asyncio
+async def test_async_post_call_failure_hook():
+    # Setup
+    logger = _ProxyDBLogger()
+
+    # Mock user_api_key_dict
+    user_api_key_dict = UserAPIKeyAuth(
+        api_key="test_api_key",
+        key_alias="test_alias",
+        user_email="test@example.com",
+        user_id="test_user_id",
+        team_id="test_team_id",
+        org_id="test_org_id",
+        team_alias="test_team_alias",
+        end_user_id="test_end_user_id",
+    )
+
+    # Mock request data
+    request_data = {
+        "model": "gpt-4",
+        "messages": [{"role": "user", "content": "Hello"}],
+        "metadata": {"original_key": "original_value"},
+        "proxy_server_request": {"request_id": "test_request_id"},
+    }
+
+    # Mock exception
+    original_exception = Exception("Test exception")
+
+    # Mock update_database function
+    with patch(
+        "litellm.proxy.proxy_server.update_database", new_callable=AsyncMock
+    ) as mock_update_database:
+        # Call the method
+        await logger.async_post_call_failure_hook(
+            request_data=request_data,
+            original_exception=original_exception,
+            user_api_key_dict=user_api_key_dict,
+        )
+
+        # Assertions
+        mock_update_database.assert_called_once()
+
+        # Check the arguments passed to update_database
+        call_args = mock_update_database.call_args[1]
+        assert call_args["token"] == "test_api_key"
+        assert call_args["response_cost"] == 0.0
+        assert call_args["user_id"] == "test_user_id"
+        assert call_args["end_user_id"] == "test_end_user_id"
+        assert call_args["team_id"] == "test_team_id"
+        assert call_args["org_id"] == "test_org_id"
+        assert call_args["completion_response"] == original_exception
+
+        # Check that metadata was properly updated
+        assert "litellm_params" in call_args["kwargs"]
+        metadata = call_args["kwargs"]["litellm_params"]["metadata"]
+        assert metadata["user_api_key"] == "test_api_key"
+        assert metadata["status"] == "failure"
+        assert "error_information" in metadata
+        assert metadata["original_key"] == "original_value"
+        assert "proxy_server_request" in metadata
+        assert metadata["proxy_server_request"]["request_id"] == "test_request_id"

tests/logging_callback_tests/test_spend_logs.py

@@ -96,6 +96,8 @@ def test_spend_logs_payload(model_id: Optional[str]):
                     },
                     "api_base": "https://openai-gpt-4-test-v-1.openai.azure.com/",
                     "caching_groups": None,
+                    "error_information": None,
+                    "status": "success",
                     "raw_request": "\n\nPOST Request Sent from LiteLLM:\ncurl -X POST \\\nhttps://openai-gpt-4-test-v-1.openai.azure.com//openai/ \\\n-H 'Authorization: *****' \\\n-d '{'model': 'chatgpt-v-2', 'messages': [{'role': 'system', 'content': 'you are a helpful assistant.\\n'}, {'role': 'user', 'content': 'bom dia'}], 'stream': False, 'max_tokens': 10, 'user': '116544810872468347480', 'extra_body': {}}'\n",
                 },
                 "model_info": {

tests/logging_callback_tests/test_standard_logging_payload.py

@@ -413,6 +413,7 @@ def test_get_error_information():
     assert result["error_code"] == "429"
     assert result["error_class"] == "RateLimitError"
     assert result["llm_provider"] == "openai"
+    assert result["error_message"] == "litellm.RateLimitError: Test error"
 
 
 def test_get_response_time():

tests/proxy_unit_tests/test_key_generate_prisma.py

@@ -507,9 +507,9 @@ def test_call_with_user_over_budget(prisma_client):
 
             # update spend using track_cost callback, make 2nd request, it should fail
             from litellm import Choices, Message, ModelResponse, Usage
-            from litellm.proxy.proxy_server import (
+            from litellm.proxy.proxy_server import _ProxyDBLogger
-                _PROXY_track_cost_callback as track_cost_callback,
+
-            )
+            proxy_db_logger = _ProxyDBLogger()
 
             resp = ModelResponse(
                 id="chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac",
@@ -526,7 +526,7 @@ def test_call_with_user_over_budget(prisma_client):
                 model="gpt-35-turbo",  # azure always has model written like this
                 usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
             )
-            await track_cost_callback(
+            await proxy_db_logger._PROXY_track_cost_callback(
                 kwargs={
                     "stream": False,
                     "litellm_params": {
@@ -604,9 +604,9 @@ def test_call_with_end_user_over_budget(prisma_client):
 
             # update spend using track_cost callback, make 2nd request, it should fail
             from litellm import Choices, Message, ModelResponse, Usage
-            from litellm.proxy.proxy_server import (
+            from litellm.proxy.proxy_server import _ProxyDBLogger
-                _PROXY_track_cost_callback as track_cost_callback,
+
-            )
+            proxy_db_logger = _ProxyDBLogger()
 
             resp = ModelResponse(
                 id="chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac",
@@ -623,7 +623,7 @@ def test_call_with_end_user_over_budget(prisma_client):
                 model="gpt-35-turbo",  # azure always has model written like this
                 usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
             )
-            await track_cost_callback(
+            await proxy_db_logger._PROXY_track_cost_callback(
                 kwargs={
                     "stream": False,
                     "litellm_params": {
@@ -711,9 +711,9 @@ def test_call_with_proxy_over_budget(prisma_client):
 
             # update spend using track_cost callback, make 2nd request, it should fail
             from litellm import Choices, Message, ModelResponse, Usage
-            from litellm.proxy.proxy_server import (
+            from litellm.proxy.proxy_server import _ProxyDBLogger
-                _PROXY_track_cost_callback as track_cost_callback,
+
-            )
+            proxy_db_logger = _ProxyDBLogger()
 
             resp = ModelResponse(
                 id="chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac",
@@ -730,7 +730,7 @@ def test_call_with_proxy_over_budget(prisma_client):
                 model="gpt-35-turbo",  # azure always has model written like this
                 usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
             )
-            await track_cost_callback(
+            await proxy_db_logger._PROXY_track_cost_callback(
                 kwargs={
                     "stream": False,
                     "litellm_params": {
@@ -802,9 +802,9 @@ def test_call_with_user_over_budget_stream(prisma_client):
 
             # update spend using track_cost callback, make 2nd request, it should fail
             from litellm import Choices, Message, ModelResponse, Usage
-            from litellm.proxy.proxy_server import (
+            from litellm.proxy.proxy_server import _ProxyDBLogger
-                _PROXY_track_cost_callback as track_cost_callback,
+
-            )
+            proxy_db_logger = _ProxyDBLogger()
 
             resp = ModelResponse(
                 id="chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac",
@@ -821,7 +821,7 @@ def test_call_with_user_over_budget_stream(prisma_client):
                 model="gpt-35-turbo",  # azure always has model written like this
                 usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
             )
-            await track_cost_callback(
+            await proxy_db_logger._PROXY_track_cost_callback(
                 kwargs={
                     "stream": True,
                     "complete_streaming_response": resp,
@@ -908,9 +908,9 @@ def test_call_with_proxy_over_budget_stream(prisma_client):
 
             # update spend using track_cost callback, make 2nd request, it should fail
             from litellm import Choices, Message, ModelResponse, Usage
-            from litellm.proxy.proxy_server import (
+            from litellm.proxy.proxy_server import _ProxyDBLogger
-                _PROXY_track_cost_callback as track_cost_callback,
+
-            )
+            proxy_db_logger = _ProxyDBLogger()
 
             resp = ModelResponse(
                 id="chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac",
@@ -927,7 +927,7 @@ def test_call_with_proxy_over_budget_stream(prisma_client):
                 model="gpt-35-turbo",  # azure always has model written like this
                 usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
             )
-            await track_cost_callback(
+            await proxy_db_logger._PROXY_track_cost_callback(
                 kwargs={
                     "stream": True,
                     "complete_streaming_response": resp,
@@ -1519,9 +1519,9 @@ def test_call_with_key_over_budget(prisma_client):
             # update spend using track_cost callback, make 2nd request, it should fail
             from litellm import Choices, Message, ModelResponse, Usage
             from litellm.caching.caching import Cache
-            from litellm.proxy.proxy_server import (
+            from litellm.proxy.proxy_server import _ProxyDBLogger
-                _PROXY_track_cost_callback as track_cost_callback,
+
-            )
+            proxy_db_logger = _ProxyDBLogger()
 
             litellm.cache = Cache()
             import time
@@ -1544,7 +1544,7 @@ def test_call_with_key_over_budget(prisma_client):
                 model="gpt-35-turbo",  # azure always has model written like this
                 usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
             )
-            await track_cost_callback(
+            await proxy_db_logger._PROXY_track_cost_callback(
                 kwargs={
                     "model": "chatgpt-v-2",
                     "stream": False,
@@ -1636,9 +1636,7 @@ def test_call_with_key_over_budget_no_cache(prisma_client):
             print("result from user auth with new key", result)
 
             # update spend using track_cost callback, make 2nd request, it should fail
-            from litellm.proxy.proxy_server import (
+            from litellm.proxy.proxy_server import _ProxyDBLogger
-                _PROXY_track_cost_callback as track_cost_callback,
-            )
             from litellm.proxy.proxy_server import user_api_key_cache
 
             user_api_key_cache.in_memory_cache.cache_dict = {}
@@ -1668,7 +1666,8 @@ def test_call_with_key_over_budget_no_cache(prisma_client):
                 model="gpt-35-turbo",  # azure always has model written like this
                 usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
             )
-            await track_cost_callback(
+            proxy_db_logger = _ProxyDBLogger()
+            await proxy_db_logger._PROXY_track_cost_callback(
                 kwargs={
                     "model": "chatgpt-v-2",
                     "stream": False,
@@ -1874,9 +1873,9 @@ async def test_call_with_key_never_over_budget(prisma_client):
         import uuid
 
         from litellm import Choices, Message, ModelResponse, Usage
-        from litellm.proxy.proxy_server import (
+        from litellm.proxy.proxy_server import _ProxyDBLogger
-            _PROXY_track_cost_callback as track_cost_callback,
+
-        )
+        proxy_db_logger = _ProxyDBLogger()
 
         request_id = f"chatcmpl-{uuid.uuid4()}"
 
@@ -1897,7 +1896,7 @@ async def test_call_with_key_never_over_budget(prisma_client):
                 prompt_tokens=210000, completion_tokens=200000, total_tokens=41000
             ),
         )
-        await track_cost_callback(
+        await proxy_db_logger._PROXY_track_cost_callback(
             kwargs={
                 "model": "chatgpt-v-2",
                 "stream": False,
@@ -1965,9 +1964,9 @@ async def test_call_with_key_over_budget_stream(prisma_client):
         import uuid
 
         from litellm import Choices, Message, ModelResponse, Usage
-        from litellm.proxy.proxy_server import (
+        from litellm.proxy.proxy_server import _ProxyDBLogger
-            _PROXY_track_cost_callback as track_cost_callback,
+
-        )
+        proxy_db_logger = _ProxyDBLogger()
 
         request_id = f"chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac{uuid.uuid4()}"
         resp = ModelResponse(
@@ -1985,7 +1984,7 @@ async def test_call_with_key_over_budget_stream(prisma_client):
             model="gpt-35-turbo",  # azure always has model written like this
             usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
         )
-        await track_cost_callback(
+        await proxy_db_logger._PROXY_track_cost_callback(
             kwargs={
                 "call_type": "acompletion",
                 "model": "sagemaker-chatgpt-v-2",
@@ -2409,9 +2408,7 @@ async def track_cost_callback_helper_fn(generated_key: str, user_id: str):
     import uuid
 
     from litellm import Choices, Message, ModelResponse, Usage
-    from litellm.proxy.proxy_server import (
+    from litellm.proxy.proxy_server import _ProxyDBLogger
-        _PROXY_track_cost_callback as track_cost_callback,
-    )
 
     request_id = f"chatcmpl-e41836bb-bb8b-4df2-8e70-8f3e160155ac{uuid.uuid4()}"
     resp = ModelResponse(
@@ -2429,7 +2426,8 @@ async def track_cost_callback_helper_fn(generated_key: str, user_id: str):
         model="gpt-35-turbo",  # azure always has model written like this
         usage=Usage(prompt_tokens=210, completion_tokens=200, total_tokens=410),
     )
-    await track_cost_callback(
+    proxy_db_logger = _ProxyDBLogger()
+    await proxy_db_logger._PROXY_track_cost_callback(
         kwargs={
             "call_type": "acompletion",
             "model": "sagemaker-chatgpt-v-2",

tests/store_model_in_db_tests/test_openai_error_handling.py

@@ -115,3 +115,97 @@ def test_missing_model_parameter_curl(curl_command):
     print("error in response", json.dumps(response, indent=4))
 
     assert "litellm.BadRequestError" in response["error"]["message"]
+
+
+@pytest.mark.asyncio
+async def test_chat_completion_bad_model_with_spend_logs():
+    """
+    Tests that Error Logs are created for failed requests
+    """
+    import json
+
+    key = generate_key_sync()
+
+    # Use httpx to make the request and capture headers
+    url = "http://0.0.0.0:4000/v1/chat/completions"
+    headers = {"Authorization": f"Bearer {key}", "Content-Type": "application/json"}
+    payload = {
+        "model": "non-existent-model",
+        "messages": [{"role": "user", "content": "Hello!"}],
+    }
+
+    with httpx.Client() as client:
+        response = client.post(url, headers=headers, json=payload)
+
+        # Extract the litellm call ID from headers
+        litellm_call_id = response.headers.get("x-litellm-call-id")
+        print(f"Status code: {response.status_code}")
+        print(f"Headers: {dict(response.headers)}")
+        print(f"LiteLLM Call ID: {litellm_call_id}")
+
+        # Parse the JSON response body
+        try:
+            response_body = response.json()
+            print(f"Error response: {json.dumps(response_body, indent=4)}")
+        except json.JSONDecodeError:
+            print(f"Could not parse response body as JSON: {response.text}")
+
+    assert (
+        litellm_call_id is not None
+    ), "Failed to get LiteLLM Call ID from response headers"
+    print("waiting for flushing error log to db....")
+    await asyncio.sleep(15)
+
+    # Now query the spend logs
+    url = "http://0.0.0.0:4000/spend/logs?request_id=" + litellm_call_id
+    headers = {"Authorization": f"Bearer sk-1234", "Content-Type": "application/json"}
+
+    with httpx.Client() as client:
+        response = client.get(
+            url,
+            headers=headers,
+        )
+
+        assert (
+            response.status_code == 200
+        ), f"Failed to get spend logs: {response.status_code}"
+
+        spend_logs = response.json()
+
+        # Print the spend logs payload
+        print(f"Spend logs response: {json.dumps(spend_logs, indent=4)}")
+
+        # Verify we have logs for the failed request
+        assert len(spend_logs) > 0, "No spend logs found"
+
+        # Check if the error is recorded in the logs
+        log_entry = spend_logs[0]  # Should be the specific log for our litellm_call_id
+
+        # Verify the structure of the log entry
+        assert log_entry["request_id"] == litellm_call_id
+        assert log_entry["model"] == "non-existent-model"
+        assert log_entry["model_group"] == "non-existent-model"
+        assert log_entry["spend"] == 0.0
+        assert log_entry["total_tokens"] == 0
+        assert log_entry["prompt_tokens"] == 0
+        assert log_entry["completion_tokens"] == 0
+
+        # Verify metadata fields
+        assert log_entry["metadata"]["status"] == "failure"
+        assert "user_api_key" in log_entry["metadata"]
+        assert "error_information" in log_entry["metadata"]
+
+        # Verify error information
+        error_info = log_entry["metadata"]["error_information"]
+        assert "traceback" in error_info
+        assert error_info["error_code"] == "400"
+        assert error_info["error_class"] == "BadRequestError"
+        assert "litellm.BadRequestError" in error_info["error_message"]
+        assert "non-existent-model" in error_info["error_message"]
+
+        # Verify request details
+        assert log_entry["cache_hit"] == "False"
+        assert log_entry["messages"]["model"] == "non-existent-model"
+        assert log_entry["messages"]["messages"][0]["role"] == "user"
+        assert log_entry["messages"]["messages"][0]["content"] == "Hello!"
+        assert log_entry["response"] == {}

ui/litellm-dashboard/src/components/view_logs/ErrorViewer.tsx

@@ -0,0 +1,164 @@
+import React from 'react';
+
+interface ErrorViewerProps {
+  errorInfo: {
+    error_class?: string;
+    error_message?: string;
+    traceback?: string;
+    llm_provider?: string;
+    error_code?: string | number;
+  };
+}
+
+export const ErrorViewer: React.FC<ErrorViewerProps> = ({ errorInfo }) => {
+  const [expandedFrames, setExpandedFrames] = React.useState<{[key: number]: boolean}>({});
+  const [allExpanded, setAllExpanded] = React.useState(false);
+  
+  // Toggle individual frame
+  const toggleFrame = (index: number) => {
+    setExpandedFrames(prev => ({
+      ...prev,
+      [index]: !prev[index]
+    }));
+  };
+  
+  // Toggle all frames
+  const toggleAllFrames = () => {
+    const newState = !allExpanded;
+    setAllExpanded(newState);
+    
+    if (tracebackFrames.length > 0) {
+      const newExpandedState: {[key: number]: boolean} = {};
+      tracebackFrames.forEach((_, idx) => {
+        newExpandedState[idx] = newState;
+      });
+      setExpandedFrames(newExpandedState);
+    }
+  };
+
+  // Parse traceback into frames
+  const parseTraceback = (traceback: string) => {
+    if (!traceback) return [];
+    
+    // Extract file paths, line numbers and code from traceback
+    const fileLineRegex = /File "([^"]+)", line (\d+)/g;
+    const matches = Array.from(traceback.matchAll(fileLineRegex));
+    
+    // Create simplified frames
+    return matches.map(match => {
+      const filePath = match[1];
+      const lineNumber = match[2];
+      const fileName = filePath.split('/').pop() || filePath;
+      
+      // Extract the context around this frame
+      const matchIndex = match.index || 0;
+      const nextMatchIndex = traceback.indexOf('File "', matchIndex + 1);
+      const frameContent = nextMatchIndex > -1 
+        ? traceback.substring(matchIndex, nextMatchIndex).trim()
+        : traceback.substring(matchIndex).trim();
+      
+      // Try to extract the code line
+      const lines = frameContent.split('\n');
+      let code = '';
+      if (lines.length > 1) {
+        code = lines[lines.length - 1].trim();
+      }
+      
+      return {
+        filePath,
+        fileName,
+        lineNumber,
+        code,
+        inFunction: frameContent.includes(' in ') 
+          ? frameContent.split(' in ')[1].split('\n')[0]
+          : ''
+      };
+    });
+  };
+
+  const tracebackFrames = errorInfo.traceback ? parseTraceback(errorInfo.traceback) : [];
+
+  return (
+    <div className="bg-white rounded-lg shadow">
+      <div className="p-4 border-b">
+        <h3 className="text-lg font-medium flex items-center text-red-600">
+          <svg className="w-5 h-5 mr-2" fill="none" stroke="currentColor" viewBox="0 0 24 24">
+            <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M12 9v2m0 4h.01m-6.938 4h13.856c1.54 0 2.502-1.667 1.732-3L13.732 4c-.77-1.333-2.694-1.333-3.464 0L3.34 16c-.77 1.333.192 3 1.732 3z" />
+          </svg>
+          Error Details
+        </h3>
+      </div>
+      
+      <div className="p-4">
+        <div className="bg-red-50 rounded-md p-4 mb-4">
+          <div className="flex">
+            <span className="text-red-800 font-medium w-20">Type:</span>
+            <span className="text-red-700">{errorInfo.error_class || "Unknown Error"}</span>
+          </div>
+          <div className="flex mt-2">
+            <span className="text-red-800 font-medium w-20">Message:</span>
+            <span className="text-red-700">{errorInfo.error_message || "Unknown error occurred"}</span>
+          </div>
+        </div>
+        
+        {errorInfo.traceback && (
+          <div className="mt-4">
+            <div className="flex justify-between items-center mb-2">
+              <h4 className="font-medium">Traceback</h4>
+              <div className="flex items-center space-x-4">
+                <button 
+                  onClick={toggleAllFrames}
+                  className="text-gray-500 hover:text-gray-700 flex items-center text-sm"
+                >
+                  {allExpanded ? "Collapse All" : "Expand All"}
+                </button>
+                <button 
+                  onClick={() => navigator.clipboard.writeText(errorInfo.traceback || "")}
+                  className="text-gray-500 hover:text-gray-700 flex items-center"
+                  title="Copy traceback"
+                >
+                  <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">
+                    <rect x="9" y="9" width="13" height="13" rx="2" ry="2"></rect>
+                    <path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"></path>
+                  </svg>
+                  <span className="ml-1">Copy</span>
+                </button>
+              </div>
+            </div>
+            
+            <div className="bg-white rounded-md border border-gray-200 overflow-hidden shadow-sm">
+              {tracebackFrames.map((frame, index) => (
+                <div key={index} className="border-b border-gray-200 last:border-b-0">
+                  <div 
+                    className="px-4 py-2 flex items-center justify-between cursor-pointer hover:bg-gray-50"
+                    onClick={() => toggleFrame(index)}
+                  >
+                    <div className="flex items-center">
+                      <span className="text-gray-400 mr-2 w-12 text-right">{frame.lineNumber}</span>
+                      <span className="text-gray-600 font-medium">{frame.fileName}</span>
+                      <span className="text-gray-500 mx-1">in</span>
+                      <span className="text-indigo-600 font-medium">{frame.inFunction || frame.fileName}</span>
+                    </div>
+                    <svg 
+                      className={`w-5 h-5 text-gray-500 transition-transform ${expandedFrames[index] ? 'transform rotate-180' : ''}`} 
+                      fill="none" 
+                      viewBox="0 0 24 24" 
+                      stroke="currentColor"
+                    >
+                      <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M19 9l-7 7-7-7" />
+                    </svg>
+                  </div>
+                  {(expandedFrames[index] || false) && frame.code && (
+                    <div className="px-12 py-2 font-mono text-sm text-gray-800 bg-gray-50 overflow-x-auto border-t border-gray-100">
+                      {frame.code}
+                    </div>
+                  )}
+                </div>
+              ))}
+            </div>
+          </div>
+        )}
+      </div>
+    </div>
+  );
+}; 

ui/litellm-dashboard/src/components/view_logs/columns.tsx

@@ -56,6 +56,24 @@ export const columns: ColumnDef<LogEntry>[] = [
     accessorKey: "startTime",
     cell: (info: any) => <TimeCell utcTime={info.getValue()} />,
   },
+  {
+    header: "Status",
+    accessorKey: "metadata.status",
+    cell: (info: any) => {
+      const status = info.getValue() || "Success";
+      const isSuccess = status.toLowerCase() !== "failure";
+      
+      return (
+        <span className={`px-2 py-1 rounded-md text-xs font-medium inline-block text-center w-16 ${
+          isSuccess 
+            ? 'bg-green-100 text-green-800' 
+            : 'bg-red-100 text-red-800'
+        }`}>
+          {isSuccess ? "Success" : "Failure"}
+        </span>
+      );
+    },
+  },
   {
     header: "Request ID",
     accessorKey: "request_id",
@@ -203,3 +221,51 @@ const formatMessage = (message: any): string => {
   }
   return String(message);
 };
+
+// Add this new component for displaying request/response with copy buttons
+export const RequestResponsePanel = ({ request, response }: { request: any; response: any }) => {
+  const requestStr = typeof request === 'object' ? JSON.stringify(request, null, 2) : String(request || '{}');
+  const responseStr = typeof response === 'object' ? JSON.stringify(response, null, 2) : String(response || '{}');
+  
+  const copyToClipboard = (text: string) => {
+    navigator.clipboard.writeText(text);
+  };
+  
+  return (
+    <div className="grid grid-cols-2 gap-4 mt-4">
+      <div className="rounded-lg border border-gray-200 bg-gray-50">
+        <div className="flex justify-between items-center p-3 border-b border-gray-200">
+          <h3 className="text-sm font-medium">Request</h3>
+          <button 
+            onClick={() => copyToClipboard(requestStr)}
+            className="p-1 hover:bg-gray-200 rounded"
+            title="Copy request"
+          >
+            <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">
+              <rect x="9" y="9" width="13" height="13" rx="2" ry="2"></rect>
+              <path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"></path>
+            </svg>
+          </button>
+        </div>
+        <pre className="p-4 overflow-auto text-xs font-mono h-64 whitespace-pre-wrap break-words">{requestStr}</pre>
+      </div>
+      
+      <div className="rounded-lg border border-gray-200 bg-gray-50">
+        <div className="flex justify-between items-center p-3 border-b border-gray-200">
+          <h3 className="text-sm font-medium">Response</h3>
+          <button 
+            onClick={() => copyToClipboard(responseStr)}
+            className="p-1 hover:bg-gray-200 rounded"
+            title="Copy response"
+          >
+            <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">
+              <rect x="9" y="9" width="13" height="13" rx="2" ry="2"></rect>
+              <path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"></path>
+            </svg>
+          </button>
+        </div>
+        <pre className="p-4 overflow-auto text-xs font-mono h-64 whitespace-pre-wrap break-words">{responseStr}</pre>
+      </div>
+    </div>
+  );
+};

ui/litellm-dashboard/src/components/view_logs/index.tsx

@@ -8,6 +8,9 @@ import { DataTable } from "./table";
 import { columns, LogEntry } from "./columns";
 import { Row } from "@tanstack/react-table";
 import { prefetchLogDetails } from "./prefetch";
+import { RequestResponsePanel } from "./columns";
+import { ErrorViewer } from './ErrorViewer';
+
 interface SpendLogsTableProps {
   accessToken: string | null;
   token: string | null;
@@ -574,146 +577,182 @@ function RequestViewer({ row }: { row: Row<LogEntry> }) {
     return input;
   };
 
+  // Extract error information from metadata if available
+  const hasError = row.original.metadata?.status === "failure";
+  const errorInfo = hasError ? row.original.metadata?.error_information : null;
+  
+  // Format the response with error details if present
+  const formattedResponse = () => {
+    if (hasError && errorInfo) {
+      return {
+        error: {
+          message: errorInfo.error_message || "An error occurred",
+          type: errorInfo.error_class || "error",
+          code: errorInfo.error_code  || "unknown",
+          param: null
+        }
+      };
+    }
+    return formatData(row.original.response);
+  };
+
   return (
     <div className="p-6 bg-gray-50 space-y-6">
       {/* Combined Info Card */}
       <div className="bg-white rounded-lg shadow">
         <div className="p-4 border-b">
-          <h3 className="text-lg font-medium ">Request Details</h3>
+          <h3 className="text-lg font-medium">Request Details</h3>
         </div>
-        <div className="space-y-2 p-4 ">
+        <div className="grid grid-cols-2 gap-4 p-4">
-          <div className="flex">
+          <div className="space-y-2">
-            <span className="font-medium w-1/3">Request ID:</span>
-            <span>{row.original.request_id}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Api Key:</span>
-            <span>{row.original.api_key}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Team ID:</span>
-            <span>{row.original.team_id}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Model:</span>
-            <span>{row.original.model}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Custom LLM Provider:</span>
-            <span>{row.original.custom_llm_provider}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Api Base:</span>
-            <span>{row.original.api_base}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Call Type:</span>
-            <span>{row.original.call_type}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Spend:</span>
-            <span>{row.original.spend}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Total Tokens:</span>
-            <span>{row.original.total_tokens}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Prompt Tokens:</span>
-            <span>{row.original.prompt_tokens}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Completion Tokens:</span>
-            <span>{row.original.completion_tokens}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Start Time:</span>
-            <span>{row.original.startTime}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">End Time:</span>
-            <span>{row.original.endTime}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Cache Hit:</span>
-            <span>{row.original.cache_hit}</span>
-          </div>
-          <div className="flex">
-            <span className="font-medium w-1/3">Cache Key:</span>
-            <span>{row.original.cache_key}</span>
-          </div>
-          {row?.original?.requester_ip_address && (
             <div className="flex">
-              <span className="font-medium w-1/3">Request IP Address:</span>
+              <span className="font-medium w-1/3">Request ID:</span>
-              <span>{row?.original?.requester_ip_address}</span>
+              <span className="font-mono text-sm">{row.original.request_id}</span>
+            </div>
+            <div className="flex">
+              <span className="font-medium w-1/3">Model:</span>
+              <span>{row.original.model}</span>
+            </div>
+            <div className="flex">
+              <span className="font-medium w-1/3">Provider:</span>
+              <span>{row.original.custom_llm_provider || "-"}</span>
+            </div>
+            <div className="flex">
+              <span className="font-medium w-1/3">Start Time:</span>
+              <span>{row.original.startTime}</span>
+            </div>
+            <div className="flex">
+              <span className="font-medium w-1/3">End Time:</span>
+              <span>{row.original.endTime}</span>
+            </div>
+          </div>
+          <div className="space-y-2">
+            <div className="flex">
+              <span className="font-medium w-1/3">Tokens:</span>
+              <span>{row.original.total_tokens} ({row.original.prompt_tokens}+{row.original.completion_tokens})</span>
+            </div>
+            <div className="flex">
+              <span className="font-medium w-1/3">Cost:</span>
+              <span>${Number(row.original.spend || 0).toFixed(6)}</span>
+            </div>
+            <div className="flex">
+              <span className="font-medium w-1/3">Cache Hit:</span>
+              <span>{row.original.cache_hit}</span>
+            </div>
+            {row?.original?.requester_ip_address && (
+              <div className="flex">
+                <span className="font-medium w-1/3">IP Address:</span>
+                <span>{row?.original?.requester_ip_address}</span>
+              </div>
+            )}
+            <div className="flex">
+              <span className="font-medium w-1/3">Status:</span>
+              <span className={`px-2 py-1 rounded-md text-xs font-medium inline-block text-center w-16 ${
+                (row.original.metadata?.status || "Success").toLowerCase() !== "failure"
+                  ? 'bg-green-100 text-green-800' 
+                  : 'bg-red-100 text-red-800'
+              }`}>
+                {(row.original.metadata?.status || "Success").toLowerCase() !== "failure" ? "Success" : "Failure"}
+              </span>
             </div>
-          )}
-        </div>
-      </div>
-
-      {/* Request Card */}
-      <div className="bg-white rounded-lg shadow">
-        <div className="flex justify-between items-center p-4 border-b">
-          <h3 className="text-lg font-medium">Request Tags</h3>
-        </div>
-        <pre className="p-4  text-wrap overflow-auto text-sm">
-          {JSON.stringify(formatData(row.original.request_tags), null, 2)}
-        </pre>
-      </div>
-
-      {/* Request Card */}
-      <div className="bg-white rounded-lg shadow">
-        <div className="flex justify-between items-center p-4 border-b">
-          <h3 className="text-lg font-medium">Request</h3>
-          {/* <div>
-            <button className="mr-2 px-3 py-1 text-sm border rounded hover:bg-gray-50">
-              Expand
-            </button>
-            <button className="px-3 py-1 text-sm border rounded hover:bg-gray-50">
-              JSON
-            </button>
-          </div> */}
-        </div>
-        <pre className="p-4  text-wrap overflow-auto text-sm">
-          {JSON.stringify(formatData(row.original.messages), null, 2)}
-        </pre>
-      </div>
-
-      {/* Response Card */}
-      <div className="bg-white rounded-lg shadow">
-        <div className="flex justify-between items-center p-4 border-b">
-          <h3 className="text-lg font-medium">Response</h3>
-          <div>
-            {/* <button className="mr-2 px-3 py-1 text-sm border rounded hover:bg-gray-50">
-              Expand
-            </button>
-            <button className="px-3 py-1 text-sm border rounded hover:bg-gray-50">
-              JSON
-            </button> */}
           </div>
         </div>
-        <pre className="p-4 text-wrap overflow-auto text-sm">
-          {JSON.stringify(formatData(row.original.response), null, 2)}
-        </pre>
       </div>
 
-      {/* Metadata Card */}
+      {/* Request/Response Panel */}
-      {row.original.metadata &&
+      <div className="grid grid-cols-2 gap-4">
-        Object.keys(row.original.metadata).length > 0 && (
+        {/* Request Side */}
-          <div className="bg-white rounded-lg shadow">
+        <div className="bg-white rounded-lg shadow">
-            <div className="flex justify-between items-center p-4 border-b">
+          <div className="flex justify-between items-center p-4 border-b">
-              <h3 className="text-lg font-medium">Metadata</h3>
+            <h3 className="text-lg font-medium">Request</h3>
-              {/* <div>
+            <button 
-                <button className="px-3 py-1 text-sm border rounded hover:bg-gray-50">
+              onClick={() => navigator.clipboard.writeText(JSON.stringify(formatData(row.original.messages), null, 2))}
-                  JSON
+              className="p-1 hover:bg-gray-200 rounded"
-                </button>
+              title="Copy request"
-              </div> */}
+            >
-            </div>
+              <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">
-            <pre className="p-4 text-wrap  overflow-auto text-sm ">
+                <rect x="9" y="9" width="13" height="13" rx="2" ry="2"></rect>
-              {JSON.stringify(row.original.metadata, null, 2)}
+                <path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"></path>
-            </pre>
+              </svg>
+            </button>
           </div>
-        )}
+          <div className="p-4 overflow-auto max-h-96">
+            <pre className="text-xs font-mono whitespace-pre-wrap break-all">{JSON.stringify(formatData(row.original.messages), null, 2)}</pre>
+          </div>
+        </div>
+
+        {/* Response Side */}
+        <div className="bg-white rounded-lg shadow">
+          <div className="flex justify-between items-center p-4 border-b">
+            <h3 className="text-lg font-medium">
+              Response
+              {hasError && (
+                <span className="ml-2 text-sm text-red-600">
+                  • HTTP code {errorInfo?.error_code || 400}
+                </span>
+              )}
+            </h3>
+            <button 
+              onClick={() => navigator.clipboard.writeText(JSON.stringify(formattedResponse(), null, 2))}
+              className="p-1 hover:bg-gray-200 rounded"
+              title="Copy response"
+            >
+              <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">
+                <rect x="9" y="9" width="13" height="13" rx="2" ry="2"></rect>
+                <path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"></path>
+              </svg>
+            </button>
+          </div>
+          <div className="p-4 overflow-auto max-h-96 bg-gray-50">
+            <pre className="text-xs font-mono whitespace-pre-wrap break-all">{JSON.stringify(formattedResponse(), null, 2)}</pre>
+          </div>
+        </div>
+      </div>
+
+
+      {/* Error Card - Only show for failures */}
+      {hasError && errorInfo && <ErrorViewer errorInfo={errorInfo} />}
+
+
+      {/* Tags Card - Only show if there are tags */}
+      {row.original.request_tags && Object.keys(row.original.request_tags).length > 0 && (
+        <div className="bg-white rounded-lg shadow">
+          <div className="flex justify-between items-center p-4 border-b">
+            <h3 className="text-lg font-medium">Request Tags</h3>
+          </div>
+          <div className="p-4">
+            <div className="flex flex-wrap gap-2">
+              {Object.entries(row.original.request_tags).map(([key, value]) => (
+                <span key={key} className="px-2 py-1 bg-gray-100 rounded-full text-xs">
+                  {key}: {String(value)}
+                </span>
+              ))}
+            </div>
+          </div>
+        </div>
+      )}
+
+      {/* Metadata Card - Only show if there's metadata */}
+      {row.original.metadata && Object.keys(row.original.metadata).length > 0 && (
+        <div className="bg-white rounded-lg shadow">
+          <div className="flex justify-between items-center p-4 border-b">
+            <h3 className="text-lg font-medium">Metadata</h3>
+            <button 
+              onClick={() => navigator.clipboard.writeText(JSON.stringify(row.original.metadata, null, 2))}
+              className="p-1 hover:bg-gray-200 rounded"
+              title="Copy metadata"
+            >
+              <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" strokeWidth="2" strokeLinecap="round" strokeLinejoin="round">
+                <rect x="9" y="9" width="13" height="13" rx="2" ry="2"></rect>
+                <path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"></path>
+              </svg>
+            </button>
+          </div>
+          <div className="p-4 overflow-auto max-h-64">
+            <pre className="text-xs font-mono whitespace-pre-wrap break-all">{JSON.stringify(row.original.metadata, null, 2)}</pre>
+          </div>
+        </div>
+      )}
     </div>
   );
 }