Rate Limits

API rate limiting and response headers

AI agent instructions

Read the Notte skill and follow its setup instructions when setting up Notte or building browser automation. This is the shared source for CLI setup, authentication, and the browser-to-code workflow.

The API rate limits a few endpoints to ensure fair usage and platform stability.

Limited endpoints

Limits are counted per API key, with a separate budget for each endpoint:

Endpoint	Description
POST /sessions/start	Start a browser session
POST /agents/start	Start an agent
POST /prompts/improve, POST /prompts/nudge	Prompt helpers

Each endpoint has a per-minute, per-hour, and per-day window. The window sizes depend on your plan; see Rate Limits for the numbers.

Rate limit exceeded (429)

When a window is exhausted, the API returns 429 with an X-Error-Class: NotteApiRateLimitExceededError header:

{
  "detail": "Rate limit exceeded: 50 per 1 minute",
  "session_id": null,
  "status": 429,
  "error": "Rate limit exceeded: 50 per 1 minute",
  "message": "Rate limit exceeded: 50 per 1 minute"
}

A 429 is also returned when you already have the maximum number of active sessions, agents, vaults, or personas for your plan. That response has X-Error-Class: NotteApiMaxActiveResourcesError and a message such as Max active sessions limit exceeded (limit: 5, current: 5). Close some of those resources, or upgrade your plan.

Rate-limit responses do not include X-RateLimit-* or Retry-After headers. Retry with exponential backoff. Retry-After is only sent on 529 responses, when no browser capacity is available (see API Error Codes).

Increasing limits

Need higher limits? Contact us to discuss enterprise options.