I get 429 "Rate limit / Too many requests" even below my tier's RPM.

Model API runs on shared capacity, so you may occasionally receive 429 errors during periods of high overall demand, even when your request rate is below your tier’s RPM limit. This is more common with high-demand models.

To mitigate this, you can:

• Add retries with exponential backoff

• Reduce request concurrency

• Temporarily fall back to another model

• Upgrade your tier for additional capacity

• Move to a Dedicated Endpoint for guaranteed capacity

If the error persists after repeated retries and pauses between requests, please contact our Support team for further assistance. Please also share your Team ID, the Model API model you’re experiencing issues with, and the relevant timestamps.