feat(microsoft-foundry/gpt-oss-20b): add new models [bot]

[models-bot]

Auto-generated by model-addition-agent for microsoft-foundry/gpt-oss-20b.

---

Note

Low Risk
Additive provider metadata only; no runtime or auth changes.

Overview
Registers gpt-oss-20b for the Microsoft Foundry provider via a new providers/microsoft-foundry/gpt-oss-20b.yaml catalog definition.

The entry marks the model as preview, provisioned, chat-only text I/O, 131k token limits, and capabilities including function calling, structured output, system messages, tool choice, and thinking: true. It also documents default max_tokens (128, up to 131072) and links to Azure AI catalog / Foundry docs.

^{Reviewed by Cursor Bugbot for commit 4f1c567. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

/test-models

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 4f1c567. Configure here.}

[cursor]

Inherited reasoning effort values unsupported

Medium Severity

The params block overrides max_tokens but does not override reasoning_effort, so the model inherits default.yaml values none, minimal, and xhigh. gpt-oss-20b only accepts low, medium, and high, so the catalog advertises settings the API rejects.

Additional Locations (1)

• providers/microsoft-foundry/gpt-oss-20b.yaml#L29-L30

 

^{Triggered by learned rule: Don't flag model-addition-agent stubs as incomplete}

^{Reviewed by Cursor Bugbot for commit 4f1c567. Configure here.}

[LordGameleo]

Approving: bot-generated model addition. Required check (Validate YAML files) is green; the gateway-test trigger failure is a CI infra issue, not a model validation failure.