Rate limits and messaging throughput

Messaging tier

How many unique customers you can deliver messages to outside a customer service window in a moving 24 hours. Since 7 October 2025 it is set per business portfolio and shared by every number in it, and it moves up automatically as quality holds and volume grows.

• Limits step up: 250, 2,000, 10,000, 100,000 and unlimited unique recipients per day.
• Only deliveries outside a customer service window count against it.
• Replies inside the service window do not.
• A quality drop no longer moves the limit back down, but it stops it growing.

Throughput

Messages per second, which is a different thing from the daily tier. A number can be well inside its tier and still hit throughput by sending a burst.

On a Coexistence number this is the limit that surprises people, because it is capped lower than on a standard Cloud API number. That ceiling is Meta's and applies to every Coexistence provider, so no provider can raise it.

Per-recipient limits

A separate cap on how many messages pass between your number and one recipient in a short window, reported as error 131056. It exists to stop one person being flooded, so it can trigger on a low-volume account.

The cause is almost always automatic: a retry loop, a webhook handler invoked twice for the same event, or a bot splitting a reply into several short messages.

Management call limits

Nothing to do with messages. Reading numbers, listing templates and checking status all count against a separate account limit, reported as error 80007.

Polling is what exhausts it. A status check every thirty seconds is thousands of calls a day against data that changes a few times a month, and webhooks remove the need for almost all of them.

What raises a limit

Quality and history, not payment. There is no plan to buy and no provider who can lift a cap on your behalf.

• Maintain a green quality rating; recipient blocks are what lower it.
• Send volume consistently rather than in bursts.
• Complete business verification, one of the paths from 250 to 2,000.
• Use at least half the current limit; increases above 2,000 are automatic.

Common mistakes

• Treating all rate limits as one. Four different limits produce four different errors.
• Polling status on a short interval. It burns the management limit on data that rarely changes.
• Expecting to pay for a higher limit. Limits move on quality and history.

Frequently asked questions

Keep reading