Runway no longer wants to be just an AI modeling company. We want to be the infrastructure layer for generative media.
On Thursday, the startup released Runway Media Router through Runway Dev, a developer platform launched earlier this month. This provides API access to Runway’s own models as well as a growing number of third-party image, video, and audio models.
Media Router is a tool that automatically selects the best image, video, or audio generation model for a request based on whether the developer prioritizes quality, speed, or cost. Model routers are becoming increasingly common in the world of large-scale language models, but Runway says this is the first one built specifically for generative media.
“This routing really fits into our overall promise of being the easiest one-stop shop for developers to integrate with all kinds of generative media models,” Anthony Maggio, Runway’s chief product officer, told TechCrunch.
The launch, shared exclusively with TechCrunch, marks another step in Runway’s evolution from an AI video startup to an enterprise infrastructure built on generative media. Through Runway Dev, customers like Adobe, Cloudflare, Celebrities, Expedia, Shutterstock, and Quora can use Runway’s APIs to embed media generation directly into their products, rather than sending users to Runway’s own apps or sites.
The launch of this router comes as the number of generated media models has exploded, making it increasingly difficult and time-consuming for developers to evaluate new releases. Developers have access to the latest released media models through the Runway Dev platform.
“Most developers don’t spend the time to really understand what each of these models does well and what’s different based on their capabilities and the different types of outputs across video, images, and audio,” Maggio says. “The unique proposition we are bringing to the table is all the intelligence about what the best model is for each different use case, and combining that with preferences to apply it to the context of the business.”
Maggio pointed out that China’s generative media model is gaining popularity. However, many companies developing their own products may not be comfortable working with Chinese-made models and may therefore prioritize U.S. model providers, he said. This priority could become more prevalent as the Trump administration considers bans and sanctions against China’s open AI models.
However, this is just one example of the settings a developer can set. Maggio says that customers are primarily interested in routing models considering the price and quality of the tokens. Token pricing became a hot topic in 2026 as companies committed to agent AI felt the pain of high token charges. Model routing for token pricing has become commonplace in the LLM world, so it’s no surprise that routing for generated media would follow suit. The launch of Media Router comes weeks after Runway replaced its unlimited subscription plan with token-based pricing, a move that drew criticism from some users.
In terms of quality, Maggio says that for generative media, determining which model provides the best quality for a given task is not as easy as it is for language models. This is where the router’s intelligence layer comes into play. This builds on the expertise Runway’s in-house creative team has developed to evaluate output across all media types, including how video models handle motion, how image models handle compositing, and how audio models handle lip sync.
Runway had already done a lot of work building the intelligence layer for its Agent product, a conversational AI creative partner Runway launched in May to help turn text prompts into fully edited multi-shot videos and marketing campaigns.
Runway Media Router takes the same routing technology Runway built for its own products and packages it for use by outside developers, Maggio said.
Runway’s current strategy reflects how fragmented and competitive today’s generative media landscape is, and how much startups need to expand and pivot to stay competitive. Runway’s last AI video model release (4.5th generation) was in December. At the time, this model topped the leaderboards, outperforming similar models from incumbents such as Google. In the same month, Runway unveiled its first global model.
Aside from upgrading to the Aleph 2.0 video editing model in May, Runway hasn’t released any new dedicated Frontier video models in recent months. (TechCrunch asked when the startup plans to release the fifth generation.) Currently, Aleph 2.0 ranks among the leading video editing models, according to Artificial Analysis, but the company’s text-to-video and image-to-video models are no longer at the top of the rankings. The top 20 rankings include models from leading companies such as Google, China’s ByteDance, and Alibaba.
Rather than asking developers to bet that a single model will continue to lead, Media Router assumes that the best model will continue to change. And you can keep Runway in the game and continue building your frontier. As the best orchestration layer, if not the best new AI model.
Anastasis Germanidis, co-founder and co-CEO of Runway, acknowledged that the startup has long been primarily known for its “end-user side,” but to get there it needed to build a full stack, including a developer platform, a suite of creative tools, and an underlying inference layer. He said he’s seeing increased interest from companies for Runway to be present in every part of their stack.
“At the core, you need good models, but orchestration becomes increasingly important as people use those models to build entire campaigns or complete multi-scene generations from those models,” Germanidis told TechCrunch. “Increasingly, we need to build an intelligence layer on top of the pure pixel model, and routers are one way that benefit can be brought to the user.”
Or, as Maggio put it more broadly, “If you look at one of the things Runway has been doing since 2018, it’s that we’re deeply focused on research and at the same time building towards the direction this space is going.”
Do you have confidential information or documents? Rebecca Beran reports on the inside world of the AI industry, from the companies shaping the future to the people affected by their decisions. Contact her securely and off the record from your non-work device on Signal at rebeccabellan.491 or by email at rebecca.bellan@techcrunch.com.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
