OpenRouter o1 or o3 not working

[andresgit]

What happened?

When trying to run using OpenRouter o1 or o3-mini-high it doesn't work and I get the following error in the log:

error: [OpenAIClient.chatCompletion] Unhandled error type Error: 400 [ { "code": "custom", "message": "Only one of \"reasoning\" and \"reasoning_effort\" may be provided", "path": [] } ]

The request seems to set both include_reasoning: true and reasoning_effort: "medium", but that doesn't work, for example this code doesn't work:

import requests
import json

api_key = "sk"
response = requests.post(
  url="https://openrouter.ai/api/v1/chat/completions",
  headers={
    "Authorization": "Bearer " + api_key,
  },
  data=json.dumps(
    {
    "model": "openai/o3-mini-high",
    "messages": [
      {
        "role": "user",
        "content": "write one sentence"
      }
    ],
    "reasoning_effort": "medium",
    "include_reasoning": True,
  }
  )
)
print(response.text.strip())

Instead the reasoning settings should be given differently: https://openrouter.ai/docs/use-cases/reasoning-tokens
This works for example:

import requests
import json

api_key = "sk"
response = requests.post(
  url="https://openrouter.ai/api/v1/chat/completions",
  headers={
    "Authorization": "Bearer " + api_key,
  },
  data=json.dumps(
    {
    "model": "openai/o3-mini-high",
    "messages": [
      {
        "role": "user",
        "content": "write one sentence"
      }
    ],
    "reasoning": {
      "effort": "medium"
    }
  }
  )
)
print(response.text.strip())

Version Information

[email protected]

Steps to Reproduce

See above

What browsers are you seeing the problem on?

No response

Relevant log output

Screenshots

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[danny-avila]

OR issue, please message them on their discord

[andresgit]

According to the documentation here https://openrouter.ai/docs/use-cases/reasoning-tokens the correct way would be {"reasoning": {"effort": "medium","exclude": False}} instead of {"reasoning_effort": "medium","include_reasoning": True}. I think this new "reasoning" parameter was just recently introduced.

[danny-avila]

Ok I see, I will try to make it work