Runway API rate limit: usage tiers, concurrency and 429s

The Runway API has no requests-per-minute rate limit. Instead, each project sits on a usage tier that caps how many generations run at once per model, how many you create in a rolling 24 hours, and how much you spend per month. Tasks over the concurrency cap wait with status THROTTLED instead of failing; going over the daily generation limit is what returns 429.

These facts come from Runway's own Usage Tiers page and the text export of its API docs, read on 2026-09-29. Runway can change them; check the page before you size a launch.

What are the Runway API usage tiers?

Limits are per model and per project. Runway's example: on Tier 3, a project can have 5 concurrent Gen-4.5 generations. All video generation models share the same concurrency limits, as do all image models. Concurrency and daily generations vary by model, so read those from the page's per-model table. The spend caps and upgrade criteria have one value per tier:

What happens when I go over a Runway API limit?

Each limit fails differently:

• Concurrency: extra tasks get status THROTTLED. They are stored on Runway's servers but not yet enqueued, and are enqueued in approximately the order you submitted them. Runway says it's safe to treat them as PENDING, but many throttled tasks can mean you're near your generation limit.
• Daily generations: counted over a 24-hour rolling window that resets continuously, not at a fixed time. Going over returns 429 Too Many Requests on the task creation request.
• Monthly spend: Runway prevents you from buying credits beyond your cap in a 30-day window, and autobilling caps automatic recharges at your remaining monthly spend.

Is the Runway API unlimited?

No. There's no requests-per-minute limit as long as you stay within your daily generations, and Runway queues work over your concurrency, so you don't need your own rate-limiting logic. But the daily cap and the monthly spend cap still apply on every tier. Runway's own example: with a concurrency of 5, you can submit 200 videos at once and they run 5 at a time in order of creation.

Runway also notes that in rare cases you may get less than your maximum concurrency, depending on system load.

How do I get higher Runway API limits?

Buy credits. When your project reaches the spend criteria for the next tier, the upgrade applies automatically with no waiting period, per the tiers page. For custom tiers, higher limits or guaranteed minimum concurrency, Runway says to file an exception request from the usage page of the developer portal. Runway's production checklist adds: if you expect a certain number of generations per day, tier up to a tier that allows that many before launch. For what those credits cost, see Runway API pricing.

How do other video APIs limit requests?

The mix of controls differs, so check each one. Sume, for example, does have a per-minute request budget per API key, and accepts jobs over its plan's concurrency as `queued` until the queue is full; see Video job concurrency and queueing.