Replace MissingEmbeddingModelError with IBM Granite default

- Replace error with ibm-granite/granite-embedding-125m-english default
- Based on issue #2418 for commercial compatibility and better UX
- Update tests to verify default fallback behavior
- Update documentation to reflect new precedence rules
- Remove unused MissingEmbeddingModelError class
- Update tip section to clarify fallback behavior

Resolves review comment to use default instead of error.

docs/source/distributions/configuration.md

@@ -705,7 +705,7 @@ Precedence rules at runtime:
 
 1. If `embedding_model` is explicitly passed in an API call, that value is used.
 2. Otherwise the value in `vector_store_config.default_embedding_model` is used.
-3. If neither is available the server will raise `MissingEmbeddingModelError` at store-creation time so mis-configuration is caught early.
+3. If neither is available the server will fall back to the system default (ibm-granite/granite-embedding-125m-english).
 
 #### Environment variables
 
@@ -721,4 +721,4 @@ export LLAMA_STACK_DEFAULT_EMBEDDING_MODEL="sentence-transformers/all-MiniLM-L6-
 llama stack run --config run.yaml
 ```
 
-> Tip: If you omit `vector_store_config` entirely you **must** either pass `embedding_model=` on every `create_vector_store` call or set `LLAMA_STACK_DEFAULT_EMBEDDING_MODEL` in the environment, otherwise the server will refuse to create a vector store.
+> Tip: If you omit `vector_store_config` entirely and don't set `LLAMA_STACK_DEFAULT_EMBEDDING_MODEL`, the system will fall back to the default `ibm-granite/granite-embedding-125m-english` model with 384 dimensions for vector store creation.

llama_stack/apis/common/vector_store_config.py

@@ -29,7 +29,7 @@ class VectorStoreConfig(BaseModel):
     default_embedding_model
         The model *id* the stack should use when an embedding model is
         required but not supplied by the API caller.  When *None* the
-        router will raise a :class:`~llama_stack.apis.common.errors.MissingEmbeddingModelError`.
+        router will fall back to the system default (ibm-granite/granite-embedding-125m-english).
     default_embedding_dimension
         Optional integer hint for vector dimension.  Routers/providers
         may validate that the chosen model emits vectors of this size.

llama_stack/core/routers/vector_io.py

@@ -11,7 +11,6 @@ from typing import Any
 from llama_stack.apis.common.content_types import (
     InterleavedContent,
 )
-from llama_stack.apis.common.errors import MissingEmbeddingModelError
 from llama_stack.apis.common.vector_store_config import VectorStoreConfig
 from llama_stack.apis.models import ModelType
 from llama_stack.apis.vector_io import (
@@ -83,7 +82,7 @@ class VectorIORouter(VectorIO):
 
         1. If *explicit_model* is provided, verify dimension (if possible) and use it.
         2. Else use the global default in ``vector_store_config``.
-        3. Else raise ``MissingEmbeddingModelError``.
+        3. Else fallback to system default (ibm-granite/granite-embedding-125m-english).
         """
 
         # 1. explicit override
@@ -106,10 +105,10 @@ class VectorIORouter(VectorIO):
         if cfg.default_embedding_model is not None:
             return cfg.default_embedding_model, cfg.default_embedding_dimension or 384
 
-        # 3. error - no default
-        raise MissingEmbeddingModelError(
-            "Failed to create vector store: No embedding model provided. Set vector_store_config.default_embedding_model or supply one in the API call."
-        )
+        # 3. fallback to system default
+        # Use IBM Granite embedding model as default for commercial compatibility
+        # See: https://github.com/meta-llama/llama-stack/issues/2418
+        return "ibm-granite/granite-embedding-125m-english", 384
 
     async def register_vector_db(
         self,

tests/unit/router/test_embedding_precedence.py

@@ -5,9 +5,6 @@
 # the root directory of this source tree.
 
 
-import pytest
-
-from llama_stack.apis.common.errors import MissingEmbeddingModelError
 from llama_stack.apis.models import ModelType
 from llama_stack.core.routers.vector_io import VectorIORouter
 
@@ -72,10 +69,11 @@ async def test_explicit_override(monkeypatch):
     monkeypatch.delenv("LLAMA_STACK_DEFAULT_EMBEDDING_MODEL", raising=False)
 
 
-async def test_error_when_no_default():
-    """Router should raise when neither explicit nor global default is available."""
+async def test_fallback_to_system_default():
+    """Router should use system default when neither explicit nor global default is available."""
 
     router = VectorIORouter(routing_table=_DummyRoutingTable())
 
-    with pytest.raises(MissingEmbeddingModelError):
-        await router._resolve_embedding_model(None)
+    model, dimension = await router._resolve_embedding_model(None)
+    assert model == "ibm-granite/granite-embedding-125m-english"
+    assert dimension == 384