Docs / Runtime
Rate limits
Rate limits apply per account by default; exceeding them returns 429.
Account type
Limit
Free / default
600 RPM · 4M TPM
Paid users
1,800 RPM · 12M TPM
Enterprise
Configured individually per contract
When you get rate limited
- The retry-after response header suggests how many seconds to wait.
- Use exponential backoff: start at 500ms, cap at 8s, at most 5 attempts.
- For batch jobs, throttle your own concurrency instead of hammering through with retries.
All keys under one account share the account-level rate limit. Per-key credit limits cap spending only, not request rate.