AI Rate Limit Calculator

OpenAI enforces two main limits per minute: RPM (requests per minute) and TPM (tokens per minute). Both limits reset every 60 seconds. If you hit either limit, subsequent requests receive a 429 error. Higher tiers offer dramatically higher limits — Tier 5 allows 10,000 RPM and 30M TPM vs Free tier's 3 RPM and 40K TPM.

Burst capacity is how many requests or tokens you can fire instantly before hitting a rate limit. Since limits reset per minute, if you send 100 requests in the first 5 seconds on Tier 1 (500 RPM limit), you'll exhaust 20% of your minute budget immediately. Burst capacity = the total limit, since OpenAI doesn't enforce sub-minute buckets.

The recommended parallel worker count is based on distributing requests evenly over 60 seconds. For example, 500 RPM with 8 requests/second means you need roughly 8 workers, each firing 1 request/second. Running too many workers causes bursty traffic and 429s; too few leaves capacity unused.

Token limits are often the binding constraint for long-context workloads. If your batch has 1,000 requests averaging 2,000 tokens each (2M tokens total) and your TPM is 200K, it'll take 10 minutes to process, not 2 minutes. This calculator shows which limit (RPM vs TPM) is the actual bottleneck.

Yes. OpenAI automatically upgrades tiers based on pre-paid credits. Tier 1 requires $5 paid, Tier 2 requires $50 paid + 7-day account age, Tier 3 requires $100 paid + 7 days, Tier 4 requires $250 paid + 14 days, Tier 5 requires $1,000 paid + 30 days. Limits reset every minute regardless of tier.