Groq `openai/gpt-oss-120b` model incompatible with structured output (`create_agent`, `ChatGroq.with_structured_output`)

[cosasha97]

Checked other resources

• This is a bug, not a usage question.
• I added a clear and descriptive title that summarizes this issue.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).
• This is not related to the langchain-community package.
• I posted a self-contained, minimal, reproducible example. A maintainer can copy it and run it AS IS.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-cli
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-perplexity
• langchain-prompty
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Example Code (Python)

from pydantic import BaseModel, Field
from typing import List, Optional
from langchain.agents import create_agent
from langchain.agents.structured_output import ToolStrategy
from langchain_groq import ChatGroq

class MovieRecommendation(BaseModel):
    title: str = Field(description="The movie title")
    genre: str = Field(description="Primary genre of the movie")
    year: Optional[int] = Field(description="Release year")
    similar_movies: List[str] = Field(description="List of 2-3 similar movies")

# Method 1: create_agent with Auto strategy (ProviderStrategy) - FAILS
agent = create_agent(model="groq:openai/gpt-oss-120b", response_format=MovieRecommendation)
result = await agent.ainvoke({"messages": [{"role": "user", "content": "Recommend a sci-fi movie"}]})

# Method 2: create_agent with ToolStrategy - FAILS
agent = create_agent(model="groq:openai/gpt-oss-120b", response_format=ToolStrategy(MovieRecommendation))
result = await agent.ainvoke({"messages": [{"role": "user", "content": "Recommend a sci-fi movie"}]})

# Method 3: ChatGroq.with_structured_output() - FAILS
llm = ChatGroq(model="openai/gpt-oss-120b")
structured_llm = llm.with_structured_output(MovieRecommendation)
result = await structured_llm.ainvoke("Recommend a sci-fi movie")

Error Message and Stack Trace (if applicable)

**Error 1: ProviderStrategy (default) - `strict` parameter not supported**


TypeError: AsyncCompletions.create() got an unexpected keyword argument 'strict'


Full traceback:

File ".../langchain_groq/chat_models.py", line 611, in _agenerate
    response = await self.async_client.create(messages=message_dicts, **params)
TypeError: AsyncCompletions.create() got an unexpected keyword argument 'strict'


**Error 2: ToolStrategy - model ignores tool calls**


BadRequestError: Error code: 400 - {'error': {'message': 'Tool choice is required, but model did not call a tool', 'type': 'invalid_request_error', 'code': 'tool_use_failed', ...}}

Description

The openai/gpt-oss-120b model on Groq is incompatible with LangChain's structured output features. Both create_agent with response_format and ChatGroq.with_structured_output() fail with different errors depending on the strategy used:

1. ProviderStrategy (default): LangChain passes strict=True to the Groq API, but the Groq SDK (groq package) does not accept this parameter for the openai/gpt-oss-120b model.

2. ToolStrategy: The model does not follow tool calling instructions and generates free-form text instead, causing a tool_use_failed error from the Groq API.

What works:

• groq:llama-3.3-70b-versatile with both ProviderStrategy (auto) and ToolStrategy works correctly
• Both structured output methods return the expected Pydantic model

What fails:

• groq:openai/gpt-oss-120b fails with both strategies (errors shown above)

Expected behavior:

Either:

1. langchain-groq should filter out the strict parameter for models that don't support it
2. Or provide a clear error message indicating the model doesn't support structured output
3. Or document which Groq models are compatible with structured output

System Info

System Information

OS: Darwin
OS Version: Darwin Kernel Version 24.5.0: Tue Apr 22 19:54:29 PDT 2025; root:xnu-11417.121.6~2/RELEASE_ARM64_T6030
Python Version: 3.12.10 (main, Apr 8 2025, 11:35:47) [Clang 16.0.0 (clang-1600.0.26.6)]

Package Information

langchain_core: 1.1.0
langchain: 1.1.0
langsmith: 0.4.49
langchain_anthropic: 1.2.0
langchain_groq: 1.1.0
langchain_openai: 1.1.0
langgraph_sdk: 0.2.10

Other Dependencies

groq: 0.36.0
openai: 1.109.1
pydantic: 2.12.5