Can't use vLLM endpoint

[brandonbiggs]

Hi. I'm trying to use AI tookit in VSCode. I can't figure out how to use my local models. I'm using vllm, not ollama, locally. My config is pretty straight forward.

version: v0.1
providers:
  - name: Custom
    models:
      - name: gpt-oss-20b
        chat_completion_url: https://example.com/v1
        headers:
          Authorization: '{"Authorization: Bearer example_token"}'
        api_model_name: gpt-oss-20b

When I call this in the playground, I see the following error:

errorObject = {"innerError":{"status":404,"headers":{},"requestID":null}} 
2025-10-14 09:19:34.973 [error] Unable to call the gpt-oss-20b inference endpoint due to 404.  Please check if the input or configuration is correct. 404 status code (no body) 

This model works and is accessible, I just can't figure out how to hook it up to AI toolkit. Any ideas would be helpful. Thanks.

[microsoft-github-policy-service]

Thank you for contacting us! Any issue or feedback from you is quite important to us. We will do our best to fully respond to your issue as soon as possible. Sometimes additional investigations may be needed, we will usually get back to you within 2 days by adding comments to this issue. Please stay tuned.