Rate limits

Every workspace has a rate limit: the number of requests per minute the API accepts from it, all API keys together. The limit depends on your plan. The Go, Plus and Max plans also cap the profiles they deliver each day. Past a limit, the API returns 429 Too Many Requests, charges 0 credits, and tells you when to retry.

Rate limit by plan

A workspace on an older credit subscription keeps the limit set with its subscription. Buying credits never lowers a limit you already have.

Your current limit is shown on the Billing page and in quotas.minuteRateLimit.limit, returned with every successful response.

How requests are counted

• One count per workspace. Every API key, the MCP server, the CLI and the dashboard draw from the same limit.
• Per calendar minute. The count starts again at the top of each minute, given in nextReset.
• One request counts once, whatever it costs. A free Check or count request counts. A search page counts once, however many profiles it returns. The request that creates an async job counts once.
• Dashboard bulk work leaves room for your API calls. A bulk enrichment or a search export started from the dashboard takes at most 80% of the limit, so your API calls always keep at least 20% of it.

These requests do not count:

• GET /v2/usage, GET /v2/flags and the Search available-fields endpoints
• Polling an async job with GET /v2/webhooks/:webhookId
• The callbacks we send to your webhookUrl
• The model turns of the Agent (its tool calls do count, like any API call)

Daily profile cap of the Go, Plus and Max plans

A Go, Plus or Max plan delivers at most 100,000 profiles per workspace per calendar day (UTC), as set out in the Terms of Service.

• What counts: each profile your plan delivers at no credit cost, through the API, the MCP server, the CLI or the dashboard. A search page of 100 profiles counts 100, and the same profile delivered twice counts twice. A 404, an empty page, a Check or a count request counts 0.
• What does not count: calls paid in credits, such as a live fetch, Find Email, Find Phone or any call your plan does not cover. They keep working once the cap is reached.
• Past the cap: the calls your plan covers return 429 with the code DAILY_EXTRACTION_LIMIT_EXCEEDED until 00:00 UTC. error.details.resetAt gives that moment.

The rate limit add-on does not raise the cap. Free trial, Pay as you go and Enterprise workspaces have no profile cap.

An Enterprise contract can set a daily request limit instead. quotas.dailyLimit shows it (null when there is none), and past it the API returns 429 with the code DAILY_LIMIT_EXCEEDED until the time in nextReset.

Limits per API key

You can cap one API key below the workspace limit, per minute or per day, for example to keep a test integration from using the whole budget. A request refused by its key's own limit does not count toward the workspace limit. See API key management.

Read your usage

The quotas object of every successful response shows where you stand in the current minute:

"minuteRateLimit": {
  "limit": 100,
  "used": 98,
  "left": 2,
  "nextReset": "2026-10-26T09:31:00.000Z"
}

left is what you can still send this minute, and nextReset is when the count starts again. GET /v2/usage returns the same object without counting toward the limit.

When you hit the limit

The API returns 429 Too Many Requests with the code RATE_LIMIT_EXCEEDED. The request costs 0 credits.

{
  "success": false,
  "data": null,
  "error": {
    "code": "RATE_LIMIT_EXCEEDED",
    "message": "Rate limit exceeded: 100 requests per 60s. Retry after 12s",
    "details": {
      "limit": 100,
      "retryAfter": 12
    }
  }
}

The response also carries two headers: Retry-After, the seconds to wait, and X-RateLimit-Reset, the moment the limit resets as a Unix timestamp. Wait the number of seconds in error.details.retryAfter, then retry. See Error handling & retries.

For a long job, such as a large file or a nightly sync, pace your requests below your limit rather than waiting for a 429: at 100 requests per minute, send one request every 600 ms at most. Read left in quotas.minuteRateLimit to slow down before you reach 0.