San Francisco, CA, Oct. 20, 2026 (GLOBE NEWSWIRE) — VideoGen, a San Francisco-based, Y Combinator-backed AI video company, today unveiled VideoGen Model 1, its first AI video generation model. Model 1 turns text prompts and images into finished, multi-shot videos up to ten minutes long. The company also opened applications for a $1 million pool of free AI credits to support selected projects by creators, filmmakers and studios, including students and amateurs. Model 1 example videos and applications for the credit program are available at videogen.io/model-1.
VideoGen Model 1 creates multi-shot AI videos up to ten minutes long from text prompts and images.
Model 1 brings creative planning, audio and video generation, and automated assembly into one workflow. Users can begin with a written brief and optional visual references. Its AI thinking harness develops the screenplay, dialogue, shots and pacing, carrying context across scenes to support consistent characters, locations, visuals, audio and narrative. Its planning system considers storytelling, narrative arcs and marketing persuasion to shape a complete video around the user’s idea.
“Lower costs and longer video generation give filmmakers room to work through scenes before they shoot and creators the ability to tell longer stories,” said Anton Koenig, CEO and Co-founder of VideoGen. “Model 1 represents a major step toward AI becoming part of everyday production, from film pre-visualization to full-length YouTube videos and ads.”
Based on internal validation, VideoGen expects Model 1 to cost more than 50% less than comparable models. Final pricing, official benchmarks and third-party validation are underway.
Model 1 supports both text-to-video and image-to-video generation. Users can begin with a text-based prompt describing the intended content, or provide up to 16 images as visual references. They can also specify a target duration and aspect ratio, allowing the production workflow to plan around the intended format.
Model 1’s underlying generation architecture uses joint audio-video attention to model visual and audio information together. It operates in latent space, using compressed representations of video and audio that are subsequently decoded into video frames and sound.
VideoGen trained and optimized Model 1 across video styles and uses, including realistic film, animation, cartoons, advertising, social media and training videos. Research techniques explored during development included style-focused post-training, low-rank adapters for text-to-video and image-to-video generation, bfloat16 inference, and experiments with model-loading and precision settings. This training and optimization work complements the AI thinking harness that coordinates creative planning and production across longer, multi-shot videos.
For videos with recurring subjects, the system maintains AI representations of characters, objects and locations and uses shared visual references across independently generated shots. The production pipeline checks for issues including incomplete dialogue and visual inconsistencies, retries selected renders within a set budget, and trims and joins the chosen clips. Longer videos are assembled from multiple generated shots.
In an internal experiment, an eight-step adapter reduced average clip-render time by 58.2% compared with the same system’s 28-step baseline. This result measures an experimental rendering configuration rather than a guaranteed speed improvement across all Model 1 outputs. VideoGen will release a technical whitepaper detailing its research approach, experiments and findings.
The $1 million credit program will support selected creative projects, with applications opening October 20. VideoGen plans to distribute the credits through the end of 2026, with individual allocations based on each project’s scope and scale. Selection will consider creative work and a viable distribution plan. Students and amateurs are eligible alongside professional creators, filmmakers and studios.
Model 1 includes safety guardrails designed to block adult and illegal content. Generated outputs carry invisible watermarks identifying their synthetic origins.
October 20 begins a limited private-partner rollout, with general availability expected to follow shortly after additional safety testing. Journalists can request early access, private demonstrations, and founder interviews by contacting press@videogen.io.
About VideoGen, Inc.
VideoGen is an AI video creation platform. Headquartered in San Francisco, VideoGen makes professional video production fast, affordable, and easy. Learn more at videogen.io.
Press Inquiries
press@videogen.io
https://videogen.io
VideoGen, Inc.
548 Market St PMB 46882
San Francisco, California 94104 US
A video accompanying this announcement is available here: https://youtube.com/watch?v=eEA6w7pX5LI
