typo

libs/core/langchain_core/language_models/chat_models.py

@@ -310,6 +310,13 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
     does not properly support streaming.
     """
 
+    model_provider: str | None = None
+    """The model provider name, e.g., 'openai', 'anthropic', etc.
+
+    Used to assign provenance on messages generated by the model, and to look up
+    model capabilities (e.g., context window sizes and feature support).
+    """
+
     output_version: str | None = Field(
         default_factory=from_env("LC_OUTPUT_VERSION", default=None)
     )
@@ -517,7 +524,9 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
                 for chunk in self._stream(input_messages, stop=stop, **kwargs):
                     if chunk.message.id is None:
                         chunk.message.id = run_id
-                    chunk.message.response_metadata = _gen_info_and_msg_metadata(chunk)
+                    chunk.message.response_metadata = _gen_info_and_msg_metadata(
+                        chunk, model_provider=self.model_provider
+                    )
                     if self.output_version == "v1":
                         # Overwrite .content with .content_blocks
                         chunk.message = _update_message_content_to_blocks(
@@ -649,7 +658,9 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
             ):
                 if chunk.message.id is None:
                     chunk.message.id = run_id
-                chunk.message.response_metadata = _gen_info_and_msg_metadata(chunk)
+                chunk.message.response_metadata = _gen_info_and_msg_metadata(
+                    chunk, model_provider=self.model_provider
+                )
                 if self.output_version == "v1":
                     # Overwrite .content with .content_blocks
                     chunk.message = _update_message_content_to_blocks(
@@ -1147,7 +1158,9 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
             index = -1
             index_type = ""
             for chunk in self._stream(messages, stop=stop, **kwargs):
-                chunk.message.response_metadata = _gen_info_and_msg_metadata(chunk)
+                chunk.message.response_metadata = _gen_info_and_msg_metadata(
+                    chunk, model_provider=self.model_provider
+                )
                 if self.output_version == "v1":
                     # Overwrite .content with .content_blocks
                     chunk.message = _update_message_content_to_blocks(
@@ -1195,6 +1208,20 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
         else:
             result = self._generate(messages, stop=stop, **kwargs)
 
+        # Add response metadata to each generation
+        for idx, generation in enumerate(result.generations):
+            if run_manager and generation.message.id is None:
+                generation.message.id = f"{LC_ID_PREFIX}-{run_manager.run_id}-{idx}"
+            generation.message.response_metadata = _gen_info_and_msg_metadata(
+                generation, model_provider=self.model_provider
+            )
+
+        if len(result.generations) == 1 and result.llm_output is not None:
+            result.generations[0].message.response_metadata = {
+                **result.llm_output,
+                **result.generations[0].message.response_metadata,
+            }
+
         if self.output_version == "v1":
             # Overwrite .content with .content_blocks
             for generation in result.generations:
@@ -1202,18 +1229,6 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
                     generation.message, "v1"
                 )
 
-        # Add response metadata to each generation
-        for idx, generation in enumerate(result.generations):
-            if run_manager and generation.message.id is None:
-                generation.message.id = f"{LC_ID_PREFIX}-{run_manager.run_id}-{idx}"
-            generation.message.response_metadata = _gen_info_and_msg_metadata(
-                generation
-            )
-        if len(result.generations) == 1 and result.llm_output is not None:
-            result.generations[0].message.response_metadata = {
-                **result.llm_output,
-                **result.generations[0].message.response_metadata,
-            }
         if check_cache and llm_cache:
             llm_cache.update(prompt, llm_string, result.generations)
         return result
@@ -1265,7 +1280,9 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
             index = -1
             index_type = ""
             async for chunk in self._astream(messages, stop=stop, **kwargs):
-                chunk.message.response_metadata = _gen_info_and_msg_metadata(chunk)
+                chunk.message.response_metadata = _gen_info_and_msg_metadata(
+                    chunk, model_provider=self.model_provider
+                )
                 if self.output_version == "v1":
                     # Overwrite .content with .content_blocks
                     chunk.message = _update_message_content_to_blocks(
@@ -1313,6 +1330,19 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
         else:
             result = await self._agenerate(messages, stop=stop, **kwargs)
 
+        # Add response metadata to each generation
+        for idx, generation in enumerate(result.generations):
+            if run_manager and generation.message.id is None:
+                generation.message.id = f"{LC_ID_PREFIX}-{run_manager.run_id}-{idx}"
+            generation.message.response_metadata = _gen_info_and_msg_metadata(
+                generation, model_provider=self.model_provider
+            )
+        if len(result.generations) == 1 and result.llm_output is not None:
+            result.generations[0].message.response_metadata = {
+                **result.llm_output,
+                **result.generations[0].message.response_metadata,
+            }
+
         if self.output_version == "v1":
             # Overwrite .content with .content_blocks
             for generation in result.generations:
@@ -1320,18 +1350,6 @@ class BaseChatModel(BaseLanguageModel[AIMessage], ABC):
                     generation.message, "v1"
                 )
 
-        # Add response metadata to each generation
-        for idx, generation in enumerate(result.generations):
-            if run_manager and generation.message.id is None:
-                generation.message.id = f"{LC_ID_PREFIX}-{run_manager.run_id}-{idx}"
-            generation.message.response_metadata = _gen_info_and_msg_metadata(
-                generation
-            )
-        if len(result.generations) == 1 and result.llm_output is not None:
-            result.generations[0].message.response_metadata = {
-                **result.llm_output,
-                **result.generations[0].message.response_metadata,
-            }
         if check_cache and llm_cache:
             await llm_cache.aupdate(prompt, llm_string, result.generations)
         return result
@@ -1719,11 +1737,15 @@ class SimpleChatModel(BaseChatModel):
 
 def _gen_info_and_msg_metadata(
     generation: ChatGeneration | ChatGenerationChunk,
+    model_provider: str | None = None,
 ) -> dict:
-    return {
+    metadata = {
         **(generation.generation_info or {}),
         **generation.message.response_metadata,
     }
+    if model_provider and "model_provider" not in metadata:
+        metadata["model_provider"] = model_provider
+    return metadata
 
 
 def _cleanup_llm_representation(serialized: Any, depth: int) -> None:

libs/core/tests/unit_tests/language_models/chat_models/test_base.py

@@ -1217,3 +1217,57 @@ def test_get_ls_params() -> None:
 
     ls_params = llm._get_ls_params(stop=["stop"])
     assert ls_params["ls_stop"] == ["stop"]
+
+
+@pytest.mark.parametrize("output_version", ["v0", "v1"])
+def test_model_provider_on_metadata(output_version: str) -> None:
+    """Test we assign model_provider to response metadata."""
+    messages = [AIMessage("hello")]
+    chunks = [AIMessageChunk(content="good"), AIMessageChunk(content="bye")]
+
+    model = _AnotherFakeChatModel(
+        responses=iter(messages),
+        chunks=iter(chunks),
+        output_version=output_version,
+        model_provider="provider_foo",
+    )
+
+    response = model.invoke("hello")
+    assert response.response_metadata["model_provider"] == "provider_foo"
+
+    response = model.invoke("hello", stream=True)
+    assert response.response_metadata["model_provider"] == "provider_foo"
+
+    model.chunks = iter([AIMessageChunk(content="good"), AIMessageChunk(content="bye")])
+    full: AIMessageChunk | None = None
+    for chunk in model.stream("hello"):
+        full = chunk if full is None else full + chunk
+    assert full is not None
+    assert full.response_metadata["model_provider"] == "provider_foo"
+
+
+@pytest.mark.parametrize("output_version", ["v0", "v1"])
+async def test_model_provider_on_metadata_async(output_version: str) -> None:
+    """Test we assign model_provider to response metadata."""
+    messages = [AIMessage("hello")]
+    chunks = [AIMessageChunk(content="good"), AIMessageChunk(content="bye")]
+
+    model = _AnotherFakeChatModel(
+        responses=iter(messages),
+        chunks=iter(chunks),
+        output_version=output_version,
+        model_provider="provider_foo",
+    )
+
+    response = await model.ainvoke("hello")
+    assert response.response_metadata["model_provider"] == "provider_foo"
+
+    response = await model.ainvoke("hello", stream=True)
+    assert response.response_metadata["model_provider"] == "provider_foo"
+
+    model.chunks = iter([AIMessageChunk(content="good"), AIMessageChunk(content="bye")])
+    full: AIMessageChunk | None = None
+    async for chunk in model.astream("hello"):
+        full = chunk if full is None else full + chunk
+    assert full is not None
+    assert full.response_metadata["model_provider"] == "provider_foo"

libs/langchain/tests/unit_tests/chat_models/test_base.py

@@ -135,6 +135,7 @@ def test_configurable() -> None:
             "model_name": "gpt-4o",
             "temperature": None,
             "model_kwargs": {},
+            "model_provider": "openai",
             "openai_api_key": SecretStr("foo"),
             "openai_api_base": None,
             "openai_organization": None,
@@ -268,6 +269,7 @@ def test_configurable_with_default() -> None:
             "betas": None,
             "default_headers": None,
             "model_kwargs": {},
+            "model_provider": None,
             "streaming": False,
             "stream_usage": True,
             "output_version": None,

libs/langchain_v1/tests/unit_tests/chat_models/test_chat_models.py

@@ -136,6 +136,7 @@ def test_configurable() -> None:
             "model_name": "gpt-4o",
             "temperature": None,
             "model_kwargs": {},
+            "model_provider": "openai",
             "openai_api_key": SecretStr("foo"),
             "openai_api_base": None,
             "openai_organization": None,
@@ -268,6 +269,7 @@ def test_configurable_with_default() -> None:
             "betas": None,
             "default_headers": None,
             "model_kwargs": {},
+            "model_provider": None,
             "streaming": False,
             "stream_usage": True,
             "output_version": None,

libs/partners/openai/langchain_openai/chat_models/base.py

@@ -550,6 +550,8 @@ class BaseChatOpenAI(BaseChatModel):
     !!! warning "Behavior changed in 0.3.35"
         Enabled for default base URL and client.
     """
+    model_provider: str | None = "openai"
+    """The model provider name (openai)."""
     max_retries: int | None = None
     """Maximum number of retries to make when generating."""
     presence_penalty: float | None = None
@@ -1034,7 +1036,6 @@ class BaseChatOpenAI(BaseChatModel):
         if usage_metadata and isinstance(message_chunk, AIMessageChunk):
             message_chunk.usage_metadata = usage_metadata
 
-        message_chunk.response_metadata["model_provider"] = "openai"
         return ChatGenerationChunk(
             message=message_chunk, generation_info=generation_info or None
         )
@@ -1398,7 +1399,6 @@ class BaseChatOpenAI(BaseChatModel):
             generations.append(gen)
         llm_output = {
             "token_usage": token_usage,
-            "model_provider": "openai",
             "model_name": response_dict.get("model", self.model_name),
             "system_fingerprint": response_dict.get("system_fingerprint", ""),
         }
@@ -4081,7 +4081,6 @@ def _construct_lc_result_from_responses_api(
     if metadata:
         response_metadata.update(metadata)
     # for compatibility with chat completion calls.
-    response_metadata["model_provider"] = "openai"
     response_metadata["model_name"] = response_metadata.get("model")
     if response.usage:
         usage_metadata = _create_usage_metadata_responses(
@@ -4277,7 +4276,6 @@ def _convert_responses_chunk_to_generation_chunk(
     tool_call_chunks: list = []
     additional_kwargs: dict = {}
     response_metadata = metadata or {}
-    response_metadata["model_provider"] = "openai"
     usage_metadata = None
     chunk_position: Literal["last"] | None = None
     id = None

libs/partners/openai/tests/unit_tests/chat_models/__snapshots__/test_azure_standard.ambr

@@ -15,6 +15,7 @@
       }),
       'max_retries': 2,
       'max_tokens': 100,
+      'model_provider': 'openai',
       'openai_api_key': dict({
         'id': list([
           'AZURE_OPENAI_API_KEY',

libs/partners/openai/tests/unit_tests/chat_models/__snapshots__/test_base_standard.ambr

@@ -11,6 +11,7 @@
       'max_retries': 2,
       'max_tokens': 100,
       'model_name': 'gpt-3.5-turbo',
+      'model_provider': 'openai',
       'openai_api_key': dict({
         'id': list([
           'OPENAI_API_KEY',

libs/partners/openai/tests/unit_tests/chat_models/__snapshots__/test_responses_standard.ambr

@@ -11,6 +11,7 @@
       'max_retries': 2,
       'max_tokens': 100,
       'model_name': 'gpt-3.5-turbo',
+      'model_provider': 'openai',
       'openai_api_key': dict({
         'id': list([
           'OPENAI_API_KEY',

libs/partners/xai/langchain_xai/chat_models.py

@@ -398,6 +398,8 @@ class ChatXAI(BaseChatOpenAI):  # type: ignore[override]
 
     model_name: str = Field(default="grok-4", alias="model")
     """Model name to use."""
+    model_provider: str | None = "xai"
+    """The model provider name (xai)."""
     xai_api_key: SecretStr | None = Field(
         alias="api_key",
         default_factory=secret_from_env("XAI_API_KEY", default=None),

libs/partners/xai/tests/unit_tests/__snapshots__/test_chat_models_standard.ambr

@@ -10,6 +10,7 @@
       'max_retries': 2,
       'max_tokens': 100,
       'model_name': 'grok-4',
+      'model_provider': 'xai',
       'request_timeout': 60.0,
       'stop': list([
       ]),