Rate limits & free models
Work within request limits and understand your daily free quota.
Requests per minute
Model API traffic is limited according to your account level. Current model API tiers are:
| Account level | Requests per minute |
|---|---|
| pro | 80 |
| max | 120 |
| ultra | 200 |
| ultra-plus | 300 |
The tools API currently uses a separate limit of 150 requests per minute per tool key. Rate limits can apply before endpoint execution. Pace workloads and bound concurrency; do not assume that adding keys increases your account allowance. Check your console for your assigned account level.
Free model policies
Free models require account verification. Daily allowances can apply at both group and model level. A text request consumes one free request unit; image requests count by requested image quantity. The Rate Limits page shows the active policy, remaining allowance, and reset time.
Policies can include limited free usage, free usage followed by paid overage, unlimited model usage subject to group rules, or unmetered usage that bypasses group quota. A model without a model-level cap may still have a group cap. Paid overage requires a configured paid price and available credits.
When you receive 429
Read the error message to distinguish the minute bucket, a daily quota, and an upstream limit. For short-lived limits, retry with exponential backoff and jitter and honor Retry-After when present. For a daily cap, wait for the reset shown in the console or use an eligible paid option. Repeated immediate retries will not restore quota.