github/langchain
1
0
Fork 0
mirror of https://github.com/hwchase17/langchain.git synced 2025-08-05 11:12:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

support service_tier

Browse Source
This commit is contained in:
Chester Curme 2025-04-24 15:40:57 -04:00
parent a7903280dd
commit 91c1c86ee1
3 changed files with 17 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
libs/partners/openai/langchain_openai/chat_models/base.py
View File

@ -538,6 +538,10 @@ class BaseChatOpenAI(BaseChatModel):
However this does not prevent a user from directly passed in the parameter during However this does not prevent a user from directly passed in the parameter during
invocation. invocation.
""" """
service_tier: Optional[str] = None
"""Latency tier for request. Options are 'auto', 'default', or 'flex'. Relevant
for users of OpenAI's scale tier service.
"""
use_responses_api: Optional[bool] = None use_responses_api: Optional[bool] = None
"""Whether to use the Responses API instead of the Chat API. """Whether to use the Responses API instead of the Chat API.
@ -655,6 +659,7 @@ class BaseChatOpenAI(BaseChatModel):
"n": self.n, "n": self.n,
"temperature": self.temperature, "temperature": self.temperature,
"reasoning_effort": self.reasoning_effort, "reasoning_effort": self.reasoning_effort,
"service_tier": self.service_tier,
} }
params = { params = {

9
libs/partners/openai/tests/integration_tests/chat_models/test_base.py
View File

@ -215,12 +215,15 @@ async def test_openai_abatch_tags(use_responses_api: bool) -> None:
assert isinstance(token.text(), str) assert isinstance(token.text(), str)
@pytest.mark.scheduled @pytest.mark.flaky(retries=3, delay=1)
def test_openai_invoke() -> None: def test_openai_invoke() -> None:
"""Test invoke tokens from ChatOpenAI.""" """Test invoke tokens from ChatOpenAI."""
llm = ChatOpenAI(max_tokens=MAX_TOKEN_COUNT) # type: ignore[call-arg] llm = ChatOpenAI(
model="o4-mini",
service_tier="flex", # Also test service_tier
)
result = llm.invoke("I'm Pickle Rick", config=dict(tags=["foo"])) result = llm.invoke("Hello", config=dict(tags=["foo"]))
assert isinstance(result.content, str) assert isinstance(result.content, str)
# assert no response headers if include_response_headers is not set # assert no response headers if include_response_headers is not set

6
libs/partners/openai/tests/unit_tests/chat_models/test_base.py
View File

@ -1732,3 +1732,9 @@ def test__construct_responses_api_input_multiple_message_types() -> None:
# assert no mutation has occurred # assert no mutation has occurred
assert messages_copy == messages assert messages_copy == messages
def test_service_tier() -> None:
llm = ChatOpenAI(model="o4-mini", service_tier="flex")
payload = llm._get_request_payload([HumanMessage("Hello")])
assert payload["service_tier"] == "flex"