Python SDK - Ollama fails

[WindowsNT]

    session_config = {
        "model": "deepseek-r1:8b",
        "streaming": True,
        "provider": {
            "type": "openai",
            "base_url": "http://localhost:11434/v1",
        },
    }

This code causes send_and_wait to wait indefinitely until python timeouts, although ollama is here and running and I can access it directly via the provided url.

[github-actions]

Thanks for reporting this issue, @WindowsNT! 👋

I've labeled this as a bug affecting the Python SDK and marked it as needs-info to help us investigate.

To help diagnose this timeout issue with Ollama, could you please provide:

1. Python SDK version (from pip show copilot)
2. Python version (from python --version)
3. Error messages or stack trace when the timeout occurs
4. Minimal reproduction code showing how you create the client and session
5. Ollama model confirmation — can you confirm curl (localhost/redacted) returns the deepseek-r1:8b` model?
6. API key setup — does the OpenAI provider require an API key? If so, have you set one (even a dummy value like "sk-dummy" may be required)?

Also, does Ollama require any authentication headers when accessed via the OpenAI-compatible endpoint?

AI generated by Issue Triage Agent

[WindowsNT]

Name: github-copilot-sdk
Version: 0.1.24

Python 3.11.6

Ollama doesn't require a key

curl works
curl http://localhost:11434/api/tags