Runway

What it is

Runway is a New York-based company that builds AI tools for making video. It was founded in 2018 by Cristóbal Valenzuela, Alejandro Matamala and Anastasis Germanidis, who had met at the ITP programme of NYU Tisch School of the Arts ^[1]. The company works mainly on generative models, which create new images and video from prompts and other inputs.

Origins

Runway began with a series of modest funding rounds. It raised a $2 million seed round in June 2018, an $8.5 million Series A in December 2020, a $35 million Series B in December 2021 and a $50 million Series C in December 2022 ^[1].

In August 2022 the company co-released Stable Diffusion, an image model, with the CompVis group at LMU Munich. The compute used for the release came from Stability AI ^[1][2]. Its release marked Runway's early work in generative media.

In June 2023 Runway raised $141 million from investors that included Google, Nvidia and Salesforce, at a reported valuation of $1.5 billion ^[3][1]. The round showed that large technology companies were investing in the company's work at that stage.

Main features

Runway's best-known work is its family of video models. Gen-1 arrived in February 2023, and Gen-2 followed later that year ^[1]. Gen-3 Alpha was released in June 2024, and Gen-4 in March 2025 ^[1]. Aleph came in July 2025, and Gen-4.5 in December 2025 ^[1][6]. Each generation extended what users could make from text and images, and the sequence shows a steady move toward longer and more capable video.

Gen-4.5 was developed entirely on NVIDIA GPUs ^[7]. Graphics processors are the chips that do the heavy calculations needed to train and run such models, and building a model entirely on one vendor's hardware means that its training depended on that vendor's chips.

Runway also runs the AI Film Festival, which grew from 300 submissions in 2023 to more than 6,000 in 2025 ^[1]. In June 2025 the festival partnered with AMC Networks ^[1]. The festival gave filmmakers who use AI tools a public place to show their work.

Studio work

In September 2024 Runway signed a deal with Lionsgate to build a custom video model. The model was to be trained on the studio's catalogue of more than 20,000 titles, and the deal was described as the first such agreement between a major studio and an AI video provider ^[6]. The arrangement shows how a generative video company can work with a film studio on a model made for that studio's own material.

Research and world models

In December 2025 Runway released its first world model, GWM-1, built on Gen-4.5 ^[8]. A world model is a system that tries to represent how a scene and its objects behave over time, so that it can simulate what happens next. This is a step beyond making a single clip, because the model has to keep track of a changing environment.

Funding

In April 2025 Runway raised a $308 million Series D led by General Atlantic. The company aimed for $300 million in annualised revenue that year ^[4]. Annualised revenue is the revenue rate over a year, so the target gave a sense of the commercial scale the company was aiming for.

On 10 February 2026 Runway raised a $315 million Series E, also led by General Atlantic, at a $5.3 billion valuation ^[5]. The money was to pre-train next-generation world models, and at that point the company had about 140 employees ^[5].

Place in the sequence

Runway's path runs from a small New York start-up to a company valued in billions, with a record of video models that moved from short clips to world models. Its partnerships, including the Lionsgate deal and its festival, connect the company to film and media, while its funding rounds show how much capital the work now requires.