London-based artificial intelligence lab Odyssey has launched a public research preview of what it calls an interactive video — a new storytelling medium generated entirely by AI in real time.
Unlike traditional video or gaming experiences, Odyssey’s new system enables users to watch, interact, and directly influence the video as it plays, with just a keystroke, swipe, or joystick movement as input.
Powered by a new world model, the system can generate and stream video frames in response to user input, creating a seamless illusion of interacting within a live digital world. Odyssey said this technology could mark the beginning of a major shift in how video is created, consumed, and experienced.
A new medium for creatives
Founded by self-driving car veterans Oliver Cameron and Jeff Hawke, Odyssey began building world models to streamline film and game production; however, through their research, they discovered an entirely new approach to storytelling.
The AI lab said this innovation isn’t meant to replace creators but to empower them.
“Interactive video . . . opens the door to entirely new forms of entertainment, where stories can be generated and explored on demand, free from the constraints and costs of traditional production,” Odyssey wrote in their blog.
A work in progress
The startup acknowledged the experience is far from perfect, noting that environments can change unexpectedly and visuals may be blurry or distorted. “The experience today feels like exploring a glitchy dream — raw, unstable, but undeniably new,” the company said.
Odyssey noted that, while the preview can be considered a “narrow distribution model,” meaning it is trained on a specific set of real-world videos to maintain stability, their team is already working on a next-generation model that promises sharper visuals, better physics, and more realistic interactions.
The early demo can stream up to 30 frames per second using powerful NVIDIA H100 GPUs, reacting to users’ actions within 40 milliseconds.
Odyssey’s big dreams
Odyssey has raised $27 million from investors, which include EQT Ventures, GV, and Air Street Capital. The startup’s board includes Ed Catmull, cofounder of Pixar and former president of Walt Disney Animation Studios.
The infrastructure cost to run this AI tool currently stands at $1 to $2 per user-hour. But Odyssey says, “This cost is decreasing fast, driven by model optimization, infrastructure investments, and tailwinds from language models.”
Zooming out, Odyssey sees a future where everything video touches evolves into interactive experiences. “Over time, we believe everything that is video today — entertainment, ads, education, training, travel, and more — will evolve into interactive video,” the company wrote.
And the Odyssey team is not alone. Other companies, including Google DeepMind, Microsoft, and Fei-Fei Li’s World Labs, are also racing to develop world models capable of building realistic, responsive digital environments.