twitter.search calls does not slow your google.search calls.
The limit applies to requests paid from your AnyAPI wallet, through REST, the SDKs, the CLI,
n8n, and MCP. Requests paid per call with x402 or MPP are not limited.
When you reach the limit
A request over the limit gets HTTP429 with error code rate_limited. The Retry-After
header gives the number of seconds until the next minute starts. Nothing runs and nothing is
charged.
run_api returns an error result with the same message, followed by a second text
block you can parse:
429 rate_limited, and the
same advice applies: wait Retry-After seconds and send the request again.
What counts
- Only new requests count toward the limit.
- Sending a request again with the
Idempotency-Keyof one that was already accepted returns that request’s result without using up the limit. While the limit is reached, that call is refused like a new one until the next minute. - Reading a request with
GET /v1/requests/{id}or MCPget_requestnever counts. It has its own pacing, described in Durable requests.
Handle a 429
- Read
Retry-After(MCP:retryAfterSeconds). - Wait that many seconds. A shorter wait is refused again, because the count has not reset yet.
- Send the same request again. It was never run, so this does not charge you twice.