feat(azure-open-ai/gpt-6-astra-2026-09-03): add new models [bot]

[models-bot]

Auto-generated by model-addition-agent for azure-open-ai/gpt-6-astra-2026-09-03.

---

Note

Low Risk
Adds a standalone model metadata YAML with no application logic or auth changes; incorrect pricing or limits would affect billing estimates only.

Overview
Adds a new Azure Open AI provider definition for gpt-6-astra-2026-09-03, enabling cost estimation and capability routing for the GA Foundry model.

The entry marks the model active with serverless provisioning, chat and responses modes, and thinking enabled via configurable reasoning_effort (default medium; temperature / top_p stripped). Pricing covers global and datazone_us with prompt-cache rates and cumulative tiered input/output/cache pricing above 272k tokens. Limits advertise a ~1.05M context window, 922k max input, 128k max output, plus text/image input and standard tool/structured-output features.

^{Reviewed by Cursor Bugbot for commit 2b282ee. 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 2b282ee. Configure here.}

[cursor]

Missing chat max_completion_tokens pair

High Severity

This fuller Azure OpenAI GPT-6 chat entry omits max_completion_tokens from params and max_tokens from removeParams. The Chat Completions API rejects max_tokens on GPT-5/GPT-6 models, so chat requests inherit the provider max_tokens default and fail.

 

^{Triggered by learned rule: Azure OpenAI GPT-5 chat needs max_completion_tokens}

^{Reviewed by Cursor Bugbot for commit 2b282ee. Configure here.}