- Media Router filters models by capability, price caps and user-defined allow or deny lists, then ranks eligible options for cost, latency or quality. [1] - Runway charges no separate routing fee; each generation is billed at the standard price of the selected model. [2] - Runway has not published benchmark results showing how often its router improves quality, speed or cost against manual model selection.
[3] Runway launched Media Router on July 23, adding an automated selection layer to its Runway Dev platform for image, video and audio generation. Developers send a request to one endpoint, and the system chooses among qualifying models according to their preferences for quality, latency and cost.[1]
The launch shifts some of the work of tracking a crowded generative-media market from developers to Runway. Its API catalog spans the company’s Gen-4.5, Aleph 2.0 and Act-Two models alongside third-party systems including Seedance, Veo, GPT Image 2 and ElevenLabs audio tools.[4]
Controls, pricing and data terms #
Users create reusable routing configurations with a maximum price per generation and model- or provider-level allow and deny lists. The router first removes models that cannot handle the request or violate a hard constraint, then scores the remaining choices. Responses identify the selected model and realized cost; a dry-run option shows what would be selected without generating media or incurring a generation charge.[1]
Runway says the router itself has no additional fee. API credits cost one cent each, and routed requests are billed at the normal rate for the chosen model. Enterprise pricing remains negotiated through sales rather than publicly listed.[2]
For enterprise accounts, Runway says its data-processing terms also cover third-party models. It says providers have contractually agreed not to train on customer inputs or outputs, while administrators can disable individual third-party models so data is not sent to them.[5]
Customers named, benchmarks absent #
TechCrunch identified Adobe, Cloudflare, ElevenLabs, Expedia, Shutterstock and Quora as Runway Dev customers, although neither Runway nor the report said those companies had deployed Media Router specifically.[3]
Runway says its quality scoring draws on evaluations by its internal creative team, including how models handle motion, composition and lip synchronization. It did not release routing accuracy, cost savings, latency comparisons or independent test results, leaving developers to validate selections through dry runs and their own production evaluations.[3]
Further sources #
[[1] Runway, “Introducing Runway Media Router,” July 23, 2026; Runway Dev model-rout… ↗](https://runway.com/news/company-news/introducing-runway-media-router)
[[2] Runway Dev, “API Pricing & Costs,” accessed July 24, 2026. ↗](https://docs.dev.runwayml.com/guides/pricing/)
[[3] TechCrunch, “Runway launches AI model router as generative media gets crowded,”… ↗](https://techcrunch.com/2026/07/23/runway-bets-on-ai-model-routing-as-generative-media-gets-crowded/)
[[4] Runway Dev, “Available AI Models,” accessed July 24, 2026. ↗](https://docs.dev.runwayml.com/guides/models/)
[[5] Runway, “Enterprise FAQ: Third-party Models in Runway,” updated April 29, 2026. ↗](https://help.runwayml.com/hc/en-us/articles/51248305153683-Enterprise-FAQ-Third-party-Models-in-Runway)
The stories that matter, in one email. Free — unsubscribe anytime.