Getting started
Rate limits and cost
Two separate allowances: how fast you may call, and how many checks you may use.
Rate limits
| Who | Limit |
|---|---|
| A request with a key | 120 calls a minute per key |
| A request without a key (the free endpoints) | 30 calls a minute per IP address |
Going over answers 429 rate_limited. These headers come on every response from a rate-limited call:
| Header | Meaning |
|---|---|
x-ratelimit-limit | The limit that applies to this call. |
x-ratelimit-remaining | Calls left in the current minute. |
x-ratelimit-reset | Seconds until the allowance resets. |
retry-after | On a 429 only: whole seconds to wait, at least 1. |
The free endpoints expose these headers to browsers, so a web page can read them.
Cost
Every successful response carries cost.checks, the number of checks the call used, so you can meter yourself. A check is one answer from one AI engine. During the beta, calls draw on your AEO Toolkit plan’s monthly checks. When a workspace has used them, calls that need checks answer 402 quota_exceeded; GET /me shows what has been used.
| Endpoint | Cost |
|---|---|
GET /ai-crawlers | Free |
POST /ai-crawlers/verify | Free |
GET /demand | Free |
GET /keywords | Free |
GET /referrals | Free |
GET /citations | Free |
POST /answers/live | 1 check per prompt per engine answered |
GET /me | Free |