Docs / Rate Limits & Quota
Rate Limits & Quota
Plan quota is metered by usage — weighted by the tokens each request processes (~1 per 10K tokens, min 1 per request, cached repeated context counts at 1/10, capped at 50 per request; failed requests are never charged — timeouts, retries, and upstream errors consume no quota), over a 5-hour window and a weekly window. Same model as Claude Code's official subscription.
5-hour window
Rolling window, auto-resets. Over the limit = temporary throttle until reset; no charge.
Weekly window
7-day fixed window, same behavior.
Concurrency
Each plan has a concurrency cap; agent clients (Claude Code / Cline) are concurrency-heavy and may hit 429.
See exact numbers on the pricing page. To see usage live in Claude Code: Usage in Claude Code.
What a 429 means
- concurrency_limit: concurrency over your plan cap — slow down / reduce concurrency, or upgrade / enable API credit billing.
- window/quota: 5h or weekly window usage exhausted — wait for reset or upgrade.
- upstream_busy: transient upstream load — usually clears shortly.
Need help? Email support@llmapi.pro · Home