Rate limits
Request limits per key, per-product limits and how to stay inside them.
Limits
| Limit | Value |
|---|---|
| Requests per key | 120 a minute |
| Question batches | 200 a day per key, up to 20 questions each |
| Words per request | 10,000 words |
| Page size for lists | Up to 100 |
| Docs playground | 10 runs a minute per visitor |
Over a limit you get 429 with a Retry-After header. Polling a job does not use your AI limits, but keep to poll_after_seconds.
Staying inside them
- Use webhooks instead of tight polling loops.
- Queue work on your side and send it at a steady rate.
- Fetch lists with
updated_sinceinstead of re-reading everything.