Rate limits

Lantfly limits how fast you can call the API and how much work an account runs at once. A limit answers 429 Too Many Requests.

Request budget

Each caller has a budget of 600 requests a minute across /api/v3 and the MCP server. The budget is counted per authenticated caller, not per token; unauthenticated requests share a budget per client address.

Some endpoints have a tighter limit of their own, counted separately for each endpoint. The API reference states them. Examples:

Endpoint	Limit
Bulk actions (…/bulk)	20 a minute
Webhook test (POST /webhooks/{id}/test)	10 a minute
Webhook redelivery (POST /webhooks/{id}/dispatches/{dispatch}/redeliver)	30 a minute
Rotate a storage signing key	20 a minute
Listening exports	20 a minute
Import listening posts by URL	10 a minute

Headers

Answers carry the standard rate-limit headers:

Header	Description
X-RateLimit-Limit	Requests allowed in the window.
X-RateLimit-Remaining	Requests left in the window.
Retry-After	On a 429: seconds to wait before retrying.
X-RateLimit-Reset	On a 429: when the window resets, as a Unix timestamp.

429 rate_limited

{ "error": { "code": "rate_limited", "message": "…" } }

Concurrent and daily runs

Your plan sets how many runs can be waiting or running at once, and may set how many runs can start in 24 hours. Starting one more answers:

Code	Status	What to do
concurrent_runs_limit_exceeded	429	Wait for a run to finish, or upgrade the plan.
daily_runs_limit_exceeded	429	Wait, or upgrade the plan.

These aren't request limits: retrying immediately won't help. Poll the runs you have, or use a webhook to learn when one finishes.

Waiting requests

Requests that hold a connection open while they wait (an MCP tool call with wait_secs, and Actor Standby) share a pool of 5 waiting requests per account. A sixth answers 429 rate_limit_exceeded with Retry-After: 5 on Standby, or a tool error on MCP. Ask for shorter waits, or call without waiting and poll.

Handle 429

1. Read Retry-After and wait that many seconds.
2. Retry with backoff and some jitter. Keep the same Idempotency-Key for writes (see Idempotency).
3. Spread bursts out: batch with bulk endpoints, page with larger per_page, and use webhooks instead of polling.

Retry on 429

import os
import random
import time

import requests

def get(url, **kwargs):
    headers = {"Authorization": f"Bearer {os.environ['LANTFLY_TOKEN']}"}
    for attempt in range(5):
        res = requests.get(url, headers=headers, timeout=30, **kwargs)
        if res.status_code != 429:
            return res
        wait = int(res.headers.get("Retry-After", 2 ** attempt))
        time.sleep(wait + random.random())
    res.raise_for_status()
    return res

Related

• Errors
• Idempotency