Growth Account Executive

Runway launches AI model router as generative media competition intensifies. Table of content. Runway has launched a new artificial intelligence model-routing tool designed to help developers select the most suitable image, video or audio model for each generation request. The Runway Media Router is available through Runway Dev, the company's developer platform that provides application programming interface access to Runway's own models and a growing selection of third-party generative media systems. The tool automatically routes requests according to priorities such as output quality, generation speed and cost. Runway says the system will reduce the time developers spend testing and comparing the increasing number of AI media models entering the market. The launch also signals a wider change in Runway's strategy. The company is expanding beyond its original position as an AI video-generation provider and attempting to become an infrastructure layer for businesses building generative media products. Media Router selects models for each request. The Runway Media Router is designed to examine a developer's request and select an appropriate model based on the preferences set for the application. A company focused on producing high-quality advertising videos may prioritize visual quality over generation speed. Another product generating large volumes of social media content may choose faster or less expensive models. Developers can adjust these priorities depending on their business requirements. Instead of manually selecting a model for every task, they can allow the router to make the decision automatically. Model routing has already become common in applications built around large language models. Companies often direct simple tasks to smaller and less expensive models while reserving more advanced systems for requests requiring stronger reasoning. Runway is applying a similar approach to generative media, where the differences between models can involve motion quality, image composition, consistency, voice generation and lip synchronization. Runway Dev provides access to multiple AI models. Runway Dev gives developers access to image, video and audio generation models through a single platform. This allows companies to add generative media features directly to their own products without sending users to Runway's website or creative application. Businesses using Runway's developer technology include Adobe, Cloudflare, ElevenLabs, Expedia, Shutterstock and Quora. These companies can use Runway's infrastructure to introduce media-generation tools while maintaining control over their own user experience. The platform is also designed to add newly released models as they become available. This could help developers experiment with emerging technology without building and maintaining separate integrations for every provider. Runway believes this centralized approach will become more valuable as the generative media market becomes increasingly fragmented. Developers face a growing number of models. The number of AI systems capable of producing images, videos and audio has increased rapidly. Major technology companies such as Google are competing alongside specialized startups and Chinese technology groups including ByteDance and Alibaba. New models frequently appear with different strengths, pricing systems and technical requirements. Evaluating these models can be difficult for developers who do not have dedicated teams for testing generative media output. A video model may perform well when producing realistic motion but struggle with text or character consistency. An image model may create strong compositions while performing poorly on complex editing instructions. Voice models can also differ in areas such as tone, timing and lip synchronization. Runway's router uses an intelligence layer developed through the company's internal evaluation of these capabilities. Its creative team examines how different models perform across a range of media formats and production tasks. The platform then combines this knowledge with the customer's selected preferences when choosing a model. Businesses can set provider and pricing preferences. Runway says developers can also establish preferences related to model providers. For example, some companies may prefer models created by American providers because of security policies, regulatory concerns or internal procurement requirements. Others may be willing to use models from a wider range of international companies when those systems offer better performance or lower prices. Cost is expected to become one of the most important routing factors. Generative AI expenses can increase quickly when applications process large numbers of requests. The issue has already affected companies using language-model applications, particularly those operating autonomous AI agents that repeatedly send requests to external models. Image and video generation can also be expensive because producing media requires substantial computing power. Automatically selecting a less costly model for routine tasks could help businesses control operating expenses. The router may direct demanding creative work to a more capable model while assigning simpler generations to a faster and more affordable alternative. Quality evaluation is more complex for AI media. Selecting the best language model can involve measuring accuracy, reasoning performance and response speed. Evaluating generative media is often more subjective. The best model can depend on the type of content being created and the desired artistic result. A model that produces realistic cinematic scenes may not be the best option for animation, product photography or video editing. Developers may also value consistency, prompt accuracy or generation speed differently depending on their application. Runway says its internal creative expertise is an important part of the Media Router. The company has spent years developing and testing tools for AI-assisted video production, image creation and editing. The routing system packages that knowledge into a service that outside developers can use. Runway previously developed similar routing technology for its AI creative agent, which helps users transform prompts into edited videos and marketing campaigns. The company is now making that underlying capability available through its developer platform. Runway expands beyond its own AI models. Runway has traditionally been known for developing advanced video-generation models. Its Gen 4.5 model gained attention after performing strongly against competing systems when it was released in December. The company also introduced a world model and later released Aleph 2.0, an upgraded video-editing system. However, competition in AI video has intensified. Models from Google, ByteDance, Alibaba and other companies now occupy leading positions across industry rankings. The Media Router allows Runway to remain central to the market even when another provider offers the strongest model for a particular task. Rather than asking customers to rely exclusively on Runway's technology, the company is offering to manage access to multiple providers and select between them. This strategy positions Runway as an orchestration platform instead of only a model developer. Orchestration becomes more important for creative AI. Generative media projects are becoming more complex as businesses move beyond producing individual images or short video clips. Companies are beginning to use AI to create complete advertising campaigns, multi-scene videos, marketing assets and interactive experiences. These projects may require several models, with each one handling a different part of the production process. One system could generate an initial image, another could animate it, and a third could produce dialogue or sound. An editing model may then combine or modify the results. Runway believes the technology responsible for coordinating these models will become as important as the individual generation systems. The company's broader platform includes its own AI research, creative tools, developer services and inference infrastructure. Media Router connects these areas by making Runway's model-selection technology available to outside businesses. Runway seeks a lasting role in generative media. Runway's latest launch reflects the changing nature of the generative AI market. Individual models can rise and fall quickly as new competitors introduce improvements in quality, speed and cost. A platform that can switch between providers may be less dependent on any single model maintaining its lead. For developers, the router could simplify access to new technology and reduce the amount of testing required before adopting a model. For Runway, the product creates another way to participate in the growth of generative media, even when customers choose models developed by competing companies. The success of the strategy will depend on whether Runway can consistently identify the best model for different tasks, provide competitive pricing and integrate new systems quickly. As the generative media industry becomes more crowded, Runway is betting that helping companies navigate the market could become just as valuable as producing the models themselves. Post comment. Be the first to post a comment!

Runway launched Media Router, a tool that automatically selects the best AI model for image, video, or audio generation based on developer priorities like quality, speed, or cost. The company says it's the first model router built specifically for generative media. The launch marks Runway's shift from AI video creator to infrastructure provider. Through its Runway Dev platform, companies including Adobe, Cloudflare, and Shutterstock can integrate media generation directly into their products. The router addresses the challenge of evaluating an exploding number of generative media models. Developers can set preferences for factors like token pricing or provider origin, with routing intelligence based on Runway's creative team's expertise. The move comes as Runway's own models no longer lead industry rankings. While its Gen 4.5 topped leaderboards in December, competitors like Google, ByteDance, and Alibaba now dominate.

A once wary Hollywood slowly warms to AI. PUBLISHED: 22 Jul 2026 at 09:45 WRITER: AFP Ask Bangkok Post anything How much did Netflix pay for InterPositive? What is the hybrid filmmaking process? Why do actors fear AI in filmmaking? NEW YORK - Once wary of artificial intelligence (AI), Hollywood is now striking up partnerships with AI start-ups and weaving the technology into productions more and more openly, all while insisting the creative process itself will remain untouched. "Everybody is still wrapping their head around it," said Mike Mosallam, COO of production company Shadowbox Studios, in an interview with Variety as recently 2025. Rattled by the technology, Hollywood initially went on the defensive. Studios worried about their intellectual property being misused, and actors feared becoming little more than raw material for AI systems and made it a central plank of their strike action in 2023. Those who did experiment ran into the technology's early limitations, six-fingered hands, mouths that did not move right. But a floodgate opened in early March, when Netflix bought a small, little-known company, fewer than 20 employees, for a hefty US$587 million. Co-founded in 2022 by producer, director and actor Ben Affleck, InterPositive is an AI platform designed to assist film and TV production. In early June, studio Lionsgate took a stake in Runway, one of the leading names in AI-generated video, and the two will now develop creative projects together. Days later, Google made a similar move, investing in independent studio A24. Recent ad campaigns and short films have shown that AI video, even when it includes human characters, has crossed a real threshold. As Runway put it in May: "In many cases, if the story is strong enough, the technology is now good enough to become invisible." According to Jamie Umpherson, Runway's chief creative officer, people are learning to use the tools better as the technology improves on a "scale that was unimaginable a couple of years ago." "There's more familiarity than there was even a year ago," he said, adding that this is freeing up creative teams to focus more on story and style. Netflix co-CEO Ted Sarandos struck a similar note when presenting the company's results in mid-July: "We're making higher quality output more quickly and efficiently than we could have using traditional methods." He said GenAI workflows have now been used on roughly 300 Netflix titles, mostly in post-production, adding: "We're leveraging GenAI for really complicated shots and sequences." 'Hybrid' Still, the streaming giant seems to be focusing mainly on the edges of production from visualising a concept early on to color-grading footage after filming wraps. Independent director and producer Jon Erwin insists the core of the craft isn't going anywhere. "Actors drive all the performances, and we use these tools to amplify the sets and the scope of the shows," he said. "But we preserve the core of the filmmaking process, which is the relationship between the actor, the director and the camera. These tools do not replace that." In April, his production company Wonder Project launched a joint venture called Innovative Dreams with Luma, a rival to Runway. Together they have developed a new process that lets actors be filmed naturally, no green screen, no motion-capture suit, and then surrounds them with a hyper-realistic AI-generated world. Erwin calls it "hybrid" or "nonlinear" filmmaking. The process is fast enough that the AI-generated backdrop can be added just minutes after a scene is shot, giving filmmakers an almost-instant preview and the chance to shoot additional takes right away if needed. The first feature film built on this approach, "Young Washington," has already grossed more than $40 million at the box office, double its production budget, starring Oscar winner Ben Kingsley alongside two Golden Globe winners. Technicians, actors and camera operators still have a place in this new ecosystem, though how many of them will be needed remains an open question. Actor Matt Damon voiced concern about this, referring to Christopher Nolan's The Odyssey, shot across three continents: "I don't think people are going to be given the resources to shoot movies that way for much longer." So far, Hollywood's traditional powerhouses, Paramount, Universal, Sony, Walt Disney and Warner Bros, have officially stayed on the sidelines of the AI shift. But Tanya Porquez, CEO of Generated Group, which has just launched a database of AI-generated video creations, thinks that won't last. "Studios will have to get on board. Not because they necessarily want to, but I think it's just the wave."