LLM API Rate Limits, Compared

Every provider throttles differently: OpenAI by spend-earned tiers, Anthropic by published per-model token budgets, Gemini by daily caps and rolling spend windows. This is the reference we wished existed — transcribed from each provider's own docs and dated, so you can tell what will actually stop your agent.

Anthropic API (Claude)
The Anthropic API rate-limits per organization across three metrics — requests per minute (RPM), input tokens per minute (ITPM), and output tokens per minute (OTPM) — with standard limits set by your usage tier (Start, Build, Scale). Exceeding any one returns a 429 with a retry-after header.
Verified 2026-07-17
Full tier tables →
OpenAI API
The OpenAI API rate-limits per organization across requests per minute/day (RPM/RPD), tokens per minute/day (TPM/TPD), images per minute, and batch-queue tokens — whichever ceiling you hit first triggers throttling. Limits grow automatically through usage tiers earned by cumulative spend.
Verified 2026-07-17
Full tier tables →
Gemini API
The Gemini API rate-limits per project across three dimensions — requests per minute (RPM), input tokens per minute (TPM), and requests per day (RPD) — plus rolling 10-minute spend caps. Exceeding any single dimension errors even if the others have headroom.
Verified 2026-07-17
Full tier tables →

The four dimensions that stop agents

RPM / RPDRequests per minute / per day. Agents trip these with parallel tool calls and retries — long before token limits.
TPM (ITPM / OTPM)Tokens per minute, sometimes split into input and output. Big contexts exhaust these with very few requests.
Usage tiersEvery provider scales limits with your spend history. New accounts start tight; limits grow automatically as you pay.
429 + retry-afterThe universal signal. Well-behaved clients back off exponentially and honor the reset headers; retry storms make it worse.

Rate limits — common questions

Why does my AI agent keep hitting rate limits?

Agents multiply every dimension at once: parallel tool calls burn requests-per-minute, long contexts burn tokens-per-minute, and always-on loops burn daily caps. Provider limits are organization-wide, so several agents share one pool — one busy agent can starve the rest.

Which LLM provider has the highest rate limits?

It depends on your tier and dimension. Anthropic publishes full per-model tables (up to 10,000 RPM and 10M input tokens/min at Scale tier); OpenAI's limits scale with cumulative spend to Tier 5; Gemini offers the most generous free tier but enforces daily request caps that hit always-on agents hardest.

How do I avoid 429 errors without paying for higher tiers?

Exponential backoff with jitter, request batching, prompt caching (on Anthropic, cache reads don't count toward input limits at all), spreading load across models, and moving non-interactive work to batch APIs. Or remove the constraint class entirely with a flat-rate gateway that paces heavy use instead of rejecting it.

Do rate limits apply per API key?

No — at every major provider, limits apply at the organization or project level. Creating more keys does not create more capacity.

The other option: no rate limits at all

Full disclosure: this reference is maintained by Standard Compute, a flat-rate gateway built on a different trade-off — no RPM, TPM, or daily caps; sustained heavy use is paced gradually instead of rejected with 429s. If your agent's problem is the limits themselves rather than tuning around them, that's the class of fix. If per-token pricing with headroom suits you, the tables above are kept honest either way.

How flat-rate unlimited works →Fixing specific 429 errors →