Rate limits

Rate limits apply to the REST API. Each limit counts requests in a fixed window, such as one calendar minute. When the window ends, the count starts again.

LimitsCopy Anchor Link

Keys from the developer console belong to an app. The app's limits are shared by all of its keys, and you can see them in the console. A new app starts with these:

Placing and removing orders (POST requests under /v1/orders) also has a limit of 40 per second. Both limits apply, so a burst of orders has to stay under both.

Keys issued before the developer console have two limits: 500 per minute for the trading routes and 240 per minute for everything else, search included.

On testnet you can call the API without a key. Requests without a key are limited per IP address: 500 per minute for the trading routes and 240 per minute for everything else. On mainnet, every request needs a key.

Read your budget from responsesCopy Anchor Link

Every response carries headers that tell you where you stand:

For example, a request to /v1/markets on testnet returns:

RateLimit-Limit: 240
RateLimit-Remaining: 239
RateLimit-Reset: 39
RateLimit-Policy: "rpm";q=240;w=60
RateLimit: "rpm";r=239;t=39

When two limits apply, as for order placement, RateLimit-Policy and RateLimit list both, such as "rps";q=40;w=1, "rpm";q=500;w=60. RateLimit-Limit, RateLimit-Remaining and RateLimit-Reset then describe whichever limit is closer to running out.

When you go overCopy Anchor Link

A request over the limit gets a 429 response, and the request is not processed:

{
  "success": false,
  "code": 429,
  "error": "too_many_requests",
  "message": "Rate limit exceeded",
  "timestamp": "2026-11-04T20:37:19.030930831Z"
}

The response has the same RateLimit-* headers. There is no Retry-After header, so use RateLimit-Reset to know how long to wait:

// Sends a request and, on a 429, waits for the rate limit window to reset before
// trying again.
async function fetchWithRetry(url: string, init: RequestInit, attempts = 3) {
  for (let attempt = 1; ; attempt++) {
    const response = await fetch(url, init);
    if (response.status !== 429 || attempt === attempts) return response;

    // seconds until the window that refused the request resets, plus a random
    // fraction so parallel clients do not retry at the same moment.
    const resetSeconds = Number(response.headers.get("RateLimit-Reset") ?? 1);
    await new Promise((resolve) =>
      setTimeout(resolve, (resetSeconds + Math.random()) * 1000),
    );
  }
}

const response = await fetchWithRetry("https://api.predict.fun/v1/markets", {
  headers: { "x-api-key": "YOUR_API_KEY" },
});
console.log(response.status, response.headers.get("RateLimit-Remaining"));

Stay under the limitsCopy Anchor Link

• Use the WebSocket API for anything that changes often, such as order books and your own orders, instead of polling the REST API.
• Watch RateLimit-Remaining and slow down before it reaches 0.
• Remove orders in batches. One request can remove up to 100 orders (see Cancel orders).