API rate limit reached
OpenClaw shows “rate limit reached” when its model provider returns a 429. The same message can come from several sources — provider RPM/TPM limits, a provider in cooldown, or no fallback configured — but it almost always traces back to an upstream provider throttling you.
OpenClaw’s heartbeats and retries are the usual culprit — they can burn a large share of a per-minute quota before you send a single message.
Point OpenClaw at Standard Compute (base URL https://api.stdcmpt.com/v1, model “standardcompute”): there’s no per-minute limit to hit, it auto-routes across providers so you never hand-configure fallbacks, and it degrades gracefully under load instead of returning 429s that stall the agent.
Background heartbeats, transient-failure retries, and parallel tool calls keep making provider requests. On a small per-minute quota, that background load alone can trigger 429s.
Configure fallbacks so a throttled provider doesn’t stall everything — or use a provider with no per-minute cap and built-in failover, so the 429 never reaches OpenClaw.