Vendor

What Runway builds, and what it distributes

Runway carries two infrastructure roles in this reference. It maintains its own video model, and on the same API it distributes ByteDance, Google, and xAI. The second role carries more weight for this dataset, because its pricing record is the only source where rival models are logged in one comparable unit.

  • Gen-4.5 is its own build
  • rival models on the same API
  • one credit equals one cent

Last checked: This is a reference page. It is re-checked against the vendor sources and updated when a new version ships. Change log

Versions

Versions Deployed Status Model families
Runway Gen-4.5 current Runway Gen

Why this record matters in a reference of model specifications

Because Runway is the one point where several rival video models are logged in a single pricing unit. Seedance 2.5, Veo 3.1, Grok Imagine 1.5, Hailuo 3, and Runway own model all sit in one table, all denominated in credits per second. The specification itself states that one credit equals one cent, so the dollar conversion is a calculation, not an estimate. When every vendor runs its own pricing unit and some publish no pricing specification at all, a distribution layer that serves them all becomes a reference source.

The figure that turns unexpected once you read the table

Runway own model, Gen-4.5, runs at twelve credits per second, equal to $0.12. The Seedance 2.0 unit Runway distributes runs thirty-six credits per second at 480p and 720p, and a hundred and fifty credits at 4K, equal to $1.50 per second. So the most expensive unit on this distribution layer costs twelve times the cheapest model Runway itself builds, and neither belongs to Runway.

Gen-4 Turbo sits on the same table at five credits per second, the cheapest video unit on the entire list. Runway outperforms its rivals on price and lands at the bottom of our quality ranking table, because that table measures output capability, not cost.

A configuration Runway distributes that Google itself does not offer

On Runway pricing record, Veo 3.1 carries two pricing specifications: forty credits per second with audio and twenty without. Google own pricing record specifies only the audio-included configuration and logs no silent row at all, because audio is always active in the Veo architecture. So this intermediary layer offers a configuration the original vendor never documented. We log this because we encountered it ourselves while building the video table.

A boundary this record does not cross

Runway also runs a consumer layer, and we record no data about it. Everything here is extracted from the API documentation and the terms-of-use record. Consumer plan pricing specifications, in-app operating ceilings, and whether any of it is reachable from Iran cannot be extracted from those two records, and we do not substitute a guess.

Documented strengths

  • the only source that logs several rival video models in one unit and one currency
  • specifies price per second of output, a directly comparable unit
  • Gen-4 Turbo at five credits per second is the cheapest video unit on the entire list
  • its OpenAPI schema is public, so real model ceilings can be extracted rather than taken from marketing claims

Known weaknesses and limits

  • Runway own model sits at the bottom of our video quality ranking table, and not by a narrow margin.
  • The pricing record does not cover everything: Kling is absent, so a dollar comparison across our four candidates stays incomplete.
  • It publishes no geographic-coverage specification and its terms cite US export control, so the Iran column on this page reads unknown.
  • Nothing here documents the consumer layer or its plans, because none of that can be extracted from the API documentation.

Technical verdict

If you want to evaluate several video models without opening separate accounts on each vendor infrastructure, the Runway API is the most sensible starting point. If your metric is peak output quality, Runway own model is not what you are provisioning there.

Frequently raised questions, with documented answers

Does Runway build its own model or distribute other vendors models?

Both. Gen-4.5, Gen-4 Turbo, and Aleph 2 are its own builds. Seedance, Veo, Grok Imagine, Hailuo, and the image and audio models on its list belong to other vendors.

What is a Runway credit worth?

Its own pricing record states that credits are purchased at $0.01 each through the developer portal. So a rate of twelve credits per second equals $0.12 per second.

Sources

  1. Runway API, available modelsvendor sourcedocs.dev.runwayml.comread on 13 August 2026
  2. Runway API pricing, per second of outputvendor sourcedocs.dev.runwayml.comread on 13 August 2026
  3. Runway terms of usevendor sourcerunway.comread on 13 August 2026