Rate limits
Hard limits per plan (no overage billing):
| Limit | Pro | Business |
|---|---|---|
| Active keys | 5 | 25 |
| Requests per minute, per key | 60 | 300 |
| Requests per minute, per account (all keys) | 120 | 600 |
| Requests per day, per account | 20,000 | 200,000 |
| Messages sent to customers per hour, per account | 60 | 600 |
| Request log kept | 30 days | 90 days |
- Weights: most operations count 1; reads with message text count 2; knowledge search and period reports count 5.
tools/listdoes not count but has its own limit of 60 per minute per key. - At most 5 messages to customers per conversation every 10 minutes (stops loops).
- REST responses carry
RateLimit-Limit,RateLimit-RemainingandRateLimit-Reset. Over the limit:429 rate_limitedwithRetry-After(seconds). Use exponential backoff. - In MCP, over the limit the tool answers with an error ("Usage limit reached; try again in N s"), so the assistant can tell the user.
- Each IP address may make up to 600 requests per minute to
/v1and/mcp.