Description
Summary
The session surfaces "Too Many Requests: Rate limit exceeded / retrying in 1s - attempt #2"
very frequently during normal use. The user expects this message to appear rarely or not at
all; instead it fires repeatedly and intrudes on the workflow.
Context
- Extension: harmoniqs.amicode 0.3.0 (darwin-arm64)
- Engine pin: v1.18.10-amicode.17 (upstream base v1.18.10), repo harmoniqs/opencode
- OS: Darwin 25.6.0 arm64
- No active run at time of report.
Expected vs actual
Expected: Rate-limit retries are infrequent or invisible; normal sessions proceed without
interruption.
Actual: The "Too Many Requests: Rate limit exceeded / retrying in 1s - attempt #2" message
surfaces very frequently during a session, indicating the retry loop fires far more often
than expected.
Diagnostics
- No run active at time of report — no run-id pointer.
- No log tail applicable.
- No absolute paths, no binary uploads.
intake: not-ready — mature per the maturity contract before picking up
suggested_path: B
diagnostics: pointer-only
upstream: anomalyco/opencode#30510 (open) — "Rate limit retries every ~1 second with no backoff or max limit"; anomalyco/opencode#33955 (open) — "Zen gateway loses provider retry timing and retries 429s early"
Activity
ariellesanford commentedon Aug 27, 2026
this is while using GLM 5.3 Flash. haven't got the same issue with other models yet. not sure if its model dependent
ariellesanford commentedon Aug 27, 2026
also got the error with Kimi k3
aarontrowbridge commentedon Oct 25, 2026
Resolved on main: the 429 retry budget (#1851) + Bedrock throttling classification (#1386), both in v0.3.9.