anomalyco / anomalyco/opencode
provider (Console Go): constant rate_limit_exceeded despite usage far below quota
@MrMushrooooom is already working on this.
Since Sep 6, 2026.
- Dominant language
- TypeScript
- Stars
- 209k
- Forks
- 27.5k
- PR merge metrics
- PR metrics pending
Description
Description
Constant [rate_limit_exceeded] errors from provider (Console Go) on every request, even though the usage dashboard shows all quotas far from exhausted (5h: 6%, weekly: 57%, monthly: 23.2%). The UI keeps scheduling automatic retries ("Retry attempt N scheduled") which also fail, making the session unusable.
Exact error:
Retry attempt 4 scheduled: Error from provider (Console Go): Upstream request failed: [rate_limit_exceeded] Rate limit exceeded. Please retry after a brief wait.
The usage dashboard at the same time shows:
- 5-hour window: 6%, resets in ~3h
- Weekly: 57%, resets in ~9h
- Monthly: 23.2%, resets in ~24 days
So either the client is hitting a different (much lower) upstream limit than the one displayed, or the displayed quota is stale/wrong, or requests are misattributed to the wrong account/bucket.
Plugins
None (default config; only a local MCP server configured, unrelated to model requests).
OpenCode version
1.18.18 (CLI). Note: the runtime diagnostics snapshot in-session reported 0.0.0-beta-19157 / channel beta — flagging in case the mismatch matters.
Steps to reproduce
- Open an opencode session using the Muse Spark model (default auth, no custom provider in
opencode.jsonc). - Send any message.
- Observe
Error from provider (Console Go): Upstream request failed: [rate_limit_exceeded]with automatic retries that also fail. - Check the usage dashboard: all windows well below 100%.
Screenshot and/or share link
No screenshot; error text pasted above verbatim.
OS
macOS 25.3.0 (darwin arm64)
Terminal
Apple_Terminal (TERM=xterm-256color, COLORTERM=truecolor)
Additional context
- Shell: /bin/zsh. Install appears to be a beta/release CLI build (1.18.18).
- Frequency: constant, not intermittent — reproduces on every request.
- Only model tested: Muse Spark (no other model tried, so unknown whether provider-wide or model-specific).
- No workaround tried yet.
- No recent config change on the user side (minimal
opencode.jsonc: single local MCP server only).
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.