LINE error 429: Too Many Requests (rate limit)

You sent requests faster than the endpoint's documented rate limit allows. The error messages table maps the body string directly: "The API rate limit has been exceeded. Try again later" means you "exceeded the rate limit for requests." But 429 on LINE is an overloaded status - the status codes table lists four distinct causes: the per-endpoint rate limit, the limit on concurrent audience operations, exceeding your free messages, and exceeding your configured cap of additional paid messages. Always read the body: quota exhaustion says You have reached your monthly limit. and needs a completely different response than throughput throttling.

The limits are per endpoint, per channel. The headline numbers from the reference: most endpoints - including reply and push - allow 2,000 requests per second; multicast allows 200 per second; broadcast and narrowcast allow 60 per hour; audience management runs at 60 per minute; the webhook settings endpoints at 1,000 per minute; rich menu create/delete at 100 per hour, the batch rich menu link at 3 per hour, and issuing short-lived tokens at 370 per second (full table at rate limits). The scope rules are documented too: the same URL with a different HTTP method is a different endpoint, parameters and IP addresses don't matter, and separate channels get independent budgets even against the same LINE Official Account.

Enforcement uses a token bucket, "rather than resetting the number of requests all at once at fixed intervals": each request spends tokens that refill gradually, short bursts can pass, and sustained overrun drains the bucket until 429s begin - then capacity returns gradually, not at the top of the minute. Notably, LINE does not document a Retry-After header or a reset timestamp, so your backoff has to be self-clocked. The push endpoint's error table also flags a subtler trigger: "a large number of messages were sent to the same user" can 429 even below the global rate.

LINE's development guidelines are blunt: "don't submit requests exceeding the rate limit." Requests refused with 429 were not processed - nothing sent, nothing charged - and are safe to retry after a delay.

HTTP/1.1 429 Too Many Requests

{
  "message": "The API rate limit has been exceeded. Try again later."
}

Encode the rate-limit table into a shared throttling layer keyed by (channel, endpoint, method), and make bulk messaging flow through multicast rather than push loops. Alert on the first sustained 429 window and on throughput approaching budget, and keep retries idempotent with retry keys. Since separate channels have separate budgets, high-volume architectures can also legitimately partition traffic across channels.