[BUG] Deep Research: Can't use reasoning_effort=high

[boxabirds]

Describe the bug
I don't seem to have access to "o1" and "o1-mini" seemed to fail so I tried claude-3.5-sonnet-xxxx and had to turn reasoning_effort=high off and set token output to 8192 and I got some… modest output.

Anyone able to get reasoning_effort=high working? For which models? I have ollama so I can use that too.

Code to reproduce the error
Just run.py in the deep research example

Error logs (if any)

    raise self._make_status_error_from_response(err.response) from None
openai.NotFoundError: Error code: 404 - {'error': {'message': 'The model `o1` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/Users/julian/.local/share/virtualenvs/open_deep_research-hTz85m-6/lib/python3.11/site-packages/litellm/main.py", line 1724, in completion
    raise e
  File "/Users/julian/.local/share/virtualenvs/open_deep_research-hTz85m-6/lib/python3.11/site-packages/litellm/main.py", line 1697, in completion
    response = openai_chat_completions.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/julian/.local/share/virtualenvs/open_deep_research-hTz85m-6/lib/python3.11/site-packages/litellm/llms/openai/openai.py", line 736, in completion
    raise OpenAIError(
litellm.llms.openai.common_utils.OpenAIError: Error code: 404 - {'error': {'message': 'The model `o1` does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}

Expected behavior
That the example would use a model that is widely available. I assume o1 is tier-restricted?

Packages version:

openinference-instrumentation-smolagents==0.1.5
smolagents==1.9.2

[aymeric-roucher]

@boxabirds it's normal not to be able to use reasoning_effotr="high" on any other model than o1. I don't know about the model being tier restricted, I've never seen this! I get no problem with using "o1" as the model_id.

[boxabirds]

Yes, when I go to openAI’s playground o1 is not a model available to me. I don’t use OpenAI much: I understand it is a tier restriction issue. I bring this up because out of the box you can’t use this unless you have o1 access and it’s not universal at least for now.

…

On Tue, 18 Feb 2025 at 15:18, Aymeric Roucher ***@***.***> wrote: @boxabirds <https://github.com/boxabirds> it's normal not to be able to use reasoning_effotr="high" on any other model than o1. I don't know about the model being tier restricted, I've never seen this! I get no problem with using "o1" as the model_id. — Reply to this email directly, view it on GitHub <#676 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABD62K646JFXVHK6L7OYUD2QNFKXAVCNFSM6AAAAABXJLE5VCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMNRWGAZDANJRG4> . You are receiving this because you were mentioned.Message ID: ***@***.***> [image: aymeric-roucher]*aymeric-roucher* left a comment (huggingface/smolagents#676) <#676 (comment)> @boxabirds <https://github.com/boxabirds> it's normal not to be able to use reasoning_effotr="high" on any other model than o1. I don't know about the model being tier restricted, I've never seen this! I get no problem with using "o1" as the model_id. — Reply to this email directly, view it on GitHub <#676 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABD62K646JFXVHK6L7OYUD2QNFKXAVCNFSM6AAAAABXJLE5VCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMNRWGAZDANJRG4> . You are receiving this because you were mentioned.Message ID: ***@***.***>