[BUG] OpenAI internal call broken

[GioPetro]

Bug Description

raise self._make_status_error_from_response(err.response) from None

openai.BadRequestError: Error code: 400 - {'error': {'message': "Unsupported parameter: 'reasoning.effort' is not supported with this model.", 'type': 'invalid_request_error', 'param': 'reasoning.effort', 'code': 'unsupported_parameter'}}

Steps to Reproduce

from graphiti_core.llm_client.openai_client import OpenAIClient
    openai_llm_config = LLMConfig(
        model=os.getenv("OPENAI_MODEL", "gpt-4o"),
        api_key=openai_api_key,
        max_tokens=128000,
        small_model="gpt-4o"
    )
    openai_llm_client = OpenAIClient(config=openai_llm_config)

    # Use OpenAI as default
    llm_client = openai_llm_client
    print("Using OpenAI client...")

    graphiti = Graphiti(NEO4J_URI, NEO4J_USER, NEO4J_PASSWORD, llm_client=llm_client)

Expected Behavior

Probably something OpenAI changed and is not handled when calling the OpenAI client. Shoulnd't be an issue whatsoever.

[GioPetro]

I see the default values being 'gpt5-mini' which makes sense now, since the rest models do not have "reasoning" attribute. Can we add a fallback without the reasoning argument to enable other openai models again? Thanks

[did-82]

I am having the same error!

[jaygokani]

Same error with the same package.

[Bundelkund]

i have the same error

[Baran121]

same error with me too am using gpt 4.1

[DonRonJohnson]

graphiti_core.llm_client.openai_base_client - ERROR - Error in generating LLM response: Error code: 400 - {'error': {'message': "Unsupported parameter: 'reasoning.effort' is not supported with this model.", 'type': 'invalid_request_error', 'param': 'reasoning.effort', 'code': 'unsupported_parameter'}}

Same this side trying to use gpt-5-mini

[claude[bot]]

@GioPetro Is this still an issue? Please confirm within 14 days or this issue will be closed.

[doriangray1945]

same error, what I need to do?