AI API rate-limit calculator
Estimate minimum batch duration, effective request throughput, and safe concurrency from documented LLM API requests-per-minute and tokens-per-minute limits.
How to use this tool
- Select a model with a documented rate-limit snapshot.
- Enter the total requests and total tokens in your batch.
- Set an expected response time to estimate a conservative concurrency level.
Worked example: Example: a 10,000-request, 2-million-token job under a 500 RPM and 500,000 TPM limit needs at least 20 minutes for its request limit alone.
Published limit500 RPM500,000 TPM
Minimum wall-clock time20 minutesLimited by requests per minute
Suggested concurrency6680% headroom at 500.0 effective RPM