Guide

Credits and limits

Every call costs 1 credit. A call that fails costs nothing. Your plan sets how fast you can go.

Credits

1 credit per call, on every route.
Failed calls are refunded. Errors and refusals are never charged.
The REST API, the MCP server and the research agent share one balance.
meta.credits_charged in each response says what the call cost.

Rate limits

The REST API is on the Pro and Enterprise plans.

Plan	Requests per minute	Requests at once
Pro	300	8
Enterprise	600	16

Over the limit

Past either limit the API answers 429 and does not charge the call. The Retry-After header says how many seconds to wait.

Code	Cause	What to do
`rate_limited`	Too many requests this minute.	Wait `Retry-After` seconds, then carry on.
`too_concurrent`	Too many requests running at once.	Let one finish, or lower your worker count to your plan's limit.

Python: respect Retry-After

import time, requests

def call(url, **kw):
    while True:
        r = requests.get(url, **kw)
        if r.status_code != 429:
            return r
        time.sleep(int(r.headers.get("Retry-After", "5")))

Need more? Plans and credits are on the pricing page.