API v0
Integration guides
Rate limits
Limits apply per API key and per endpoint. Expert capacity is handled by queueing, not errors.
Each API key has a request rate limit per endpoint, covering POST /questions, GET /questions/{request_id} and GET /answers/search. Calls over the limit return 429 rate_limited with a Retry-After header. If you need higher limits, talk to us.
Calls versus panels
Rate limits govern API calls, not expert panels. Submissions above expert capacity are queued rather than rejected, and the eta reflects the queue. Open requests per workspace are capped; beyond the cap, new submissions are queued too.
Polling politely
Polling a pending request every few minutes is plenty, because answers arrive over hours. Better still, use webhooks and fetch the request when an event arrives.
Spend limits are separate
Limits on answers per request, per day and per month are set in Usage and billing and return 403 limit_exceeded.
Next
MCP integration →