Runway Launches Media Router to Auto-Select Best AI Model for Image, Video, Audio Generation

Runway launched Media Router on Thursday, a tool that automatically selects the best image, video, or audio generation model for a given request based on whether a developer prioritizes quality, speed, or cost. The company says it is the first model router built specifically for generative media, according to Runway.

The tool ships through Runway Dev, the developer platform Runway launched earlier this month that provides API access to a growing roster of third-party and in-house image, video, and audio models. Customers including Adobe, Cloudflare, ElevenLabs, Expedia, Shutterstock, and Quora already use Runway Dev to build media generation directly into their own products rather than routing users to Runway's own app.

"The routing really fits into that overall promise of being the easiest one-stop shop for developers to integrate with any type of generative media model," Runway chief product officer Anthony Maggio told TechCrunch.

Why now

Model routing is already standard practice among large language model providers, where developers commonly switch between models to manage token costs and latency. Runway is betting the same logic applies to generative media as the number of image, video, and audio models has grown rapidly, making it harder for developers to track which model performs best for a given use case.

"Most developers are not spending the time to really understand the capabilities of each of these models and where they excel or differ based on various types of outputs across video, image, and audio," Maggio said.

The router's decision-making draws on evaluation criteria Runway's in-house creative team developed for judging output quality — how video models handle motion, how image models handle composition, how voice models handle lip syncing. That same intelligence layer was originally built for Runway's conversational agent product, launched in May, which turns text prompts into multi-shot videos and marketing campaigns.

Developers can also set preferences unrelated to output quality. Maggio noted that Chinese generative media models are gaining popularity, but some businesses may prefer to exclude them given the Trump administration's exploration of bans and sanctions against Chinese open AI models. Token pricing and quality remain the two preferences customers set most often, according to Maggio.

Context: Runway's shifting position

The launch comes as Runway's own frontier models have lost ground on leaderboards. Gen-4.5, released in December, topped rankings at launch, but Runway's text-to-video and image-to-video models no longer lead Artificial Analysis rankings, with Google, ByteDance, and Alibaba models now occupying top spots. Runway's video editing model, Aleph 2.0, updated in May, still ranks among the leading editing tools. TechCrunch has asked Runway when it plans to release Gen-5; the company has not disclosed a timeline.

The Media Router also follows Runway's replacement of unlimited subscription plans with token-based pricing weeks earlier — a change that drew criticism from some users.

Co-founder and co-CEO Anastasis Germanidis said Runway has built a full stack beyond its consumer app: a developer platform, creative tool suite, and inference layer. "The orchestration increasingly matters a lot because people are building entire campaigns with those models," he said.

What this means

Runway's pivot signals a hedge: if it can't guarantee its own models stay at the frontier of generative media, it can still position itself as the layer developers route through regardless of which lab wins any given benchmark cycle. That's a defensible strategy in a fragmented market, but it also depends on Runway's routing intelligence being genuinely better than developers picking models themselves — a claim not yet independently verified. Pricing details for Media Router access were not disclosed.