> ## Documentation Index > Fetch the complete documentation index at: https://docs.meowmail.in/llms.txt > Use this file to discover all available pages before exploring further. # Rate limits > Two tiers, two windows, and the headers that let you self-throttle. | Tier | Per minute | Per day | Concurrent waits | | ------------------ | ---------- | ------- | ---------------- | | Anonymous (per IP) | 60 | 1,000 | — | | Free key (per key) | 300 | 10,000 | 5 | Both windows apply. Staying under the per-minute limit does not exempt you from the daily quota. ## Read the headers instead of guessing Every response carries the current state, so you never have to discover a limit by being refused: ``` ratelimit-limit: 300 ratelimit-remaining: 297 ratelimit-reset: 60 ``` A `429` additionally carries `retry-after` in seconds. Honour it — retrying sooner just burns quota against a closed window. ## Why anonymous callers should still get a key Anonymous requests are bucketed **by IP**. CI runners, serverless functions, and corporate NATs share egress addresses, which means unrelated users throttle each other and you cannot tell your own traffic from theirs. A key gives you a private bucket at 5× the rate. That is the honest reason to get one, and it is why we do not need to force it. Service discovery (`GET /api/v1`), the OpenAPI spec, and `GET /api/v1/health` sit **outside** the developer quota. A caller who has just been rate-limited must still be able to read the endpoint that explains why, and a monitoring system polling health should not exhaust a budget shared with its neighbours. ## Need more? Limits are per-key, so raising yours is a one-line change on our side. Email [dev.jaythorat@gmail.com](mailto:dev.jaythorat@gmail.com?subject=MeowMail%20API) with your key prefix and what you are building. Tell us what you need rather than working around the limit by minting keys in bulk — that reads as abuse from our side and gets the whole IP revoked.