Uh oh!
There was an error while loading. Please reload this page.
improvement(openai): migrate to responses api - #3135
Conversation
The latest updates on your projects. Learn more about Vercel for GitHub. |
Sg312
commented
Feb 4, 2026
@cursor review |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Sg312
commented
Feb 4, 2026
@cursor review |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Sg312
commented
Feb 4, 2026
@cursor review |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Sg312
commented
Feb 4, 2026
@cursor review |
Uh oh!
There was an error while loading. Please reload this page.
Sg312
commented
Feb 4, 2026
@cursor review |
Sg312
commented
Feb 4, 2026
@cursor review |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Sg312
commented
Feb 4, 2026
@cursor review |
Uh oh!
There was an error while loading. Please reload this page.
Sg312
commented
Feb 4, 2026
@cursor review |
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
* Migrate openai to use responses api * Consolidate azure * Fix streaming * Bug fixes * Bug fixes * Fix responseformat * Refactor * Fix bugs * Fix * Fix azure openai response format with tool calls * Fixes * Fixes * Fix temp
…path `/v1/responses` posted through the OpenAI SDK until 1933e1a (#3135) moved it onto raw `fetch`, which silently dropped the SDK's `maxRetries: 2`. The 16 other providers that construct an SDK client still retry; only openai and azure-openai, which share this core, retried nothing. Restores 2 retries (3 attempts) on 408/409/429/5xx using `backoffWithJitter` and `parseRetryAfter`, preferring OpenAI's `retry-after-ms` over `Retry-After`. The loop sits in the shared request helper so the streaming paths are covered too — a refused request yields no body, so no stream bytes were consumed and no response was created server-side. Aborts, other 4xx, and the body-read deadline stay non-retryable. `/v1/responses` ignores `Idempotency-Key`, so a retry after a response already exists would generate and bill a second one; the body stall is exactly that case.
Summary
Migrate openai provider to use responses api
Type of Change
Testing
Manual with @waleedlatif1
Checklist
- Code follows project style guidelines
- Self-reviewed my changes
- Tests added/updated and passing
- No new warnings introduced
- I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)
Cursor Bugbot found 1 potential issue for commit 95bdc33