Usage
Rate limits
Limits that protect capacity, and how to work within them.
Requests are limited per workspace and per model to keep capacity fair. Limits are defined by plan and agreement.
When you hit a limit
The API returns HTTP 429. Wait and retry with exponential backoff.
Reducing pressure
- Batch work where latency allows.
- Cache responses for repeated inputs.
- Spread scheduled jobs over time.
- Contact the team about higher limits for production traffic.