Upstream rate limit error from Alibaba model provider — "Request rate increased too quickly"

[runboktu]

Question

Title: Upstream rate limit error from Alibaba model provider — "Request rate increased too quickly"
Body:
Environment

• OpenCode version: (run opencode --version)
• Model: qwen3.6-plus-free (Alibaba / DashScope)
• Platform: macOS (darwin)
• Node.js version: (run node --version)
  Error Message
  Upstream error from Alibaba: Request rate increased too quickly. To ensure system stability, please adjust your client logic to scale requests more smoothly over time.
  What Happens
  When using OpenCode with the Alibaba Qwen model (qwen3.6-plus-free), the upstream API returns a rate-limit error. This causes the session to fail/break, especially during:

[github-actions]

This issue might be a duplicate of existing issues. Please check:

• Upstream error #21217: Same upstream Alibaba error message ("Request rate increased too quickly") with Qwen model
• Qwen 3.6 plus (free) #21164: Same exact error with Qwen 3.6 plus (free) model, same question about reducing request rate
• [FEATURE]: Add automatic retry with backoff for Alibaba/DashScope rate limit errors #21140: Feature request for automatic retry with backoff specifically for Alibaba/DashScope rate limit errors