News
VideoGen Model 1 Promises Multi-Shot Videos Up to Ten Minutes From 16 References, and a $1 Million Credit Pool
· RenderBob team
VideoGen announced Model 1 on 20 October. The company says it plans and assembles multi-shot videos up to ten minutes long from a prompt and up to 16 images.
VideoGen announced Model 1 on 20 October, a system it says plans and assembles multi-shot videos up to ten minutes long. It is in a limited partner rollout. The claims below come from the company's GlobeNewswire release that day.
What the release says
VideoGen, a San Francisco company backed by Y Combinator, says Model 1 turns a text prompt and up to 16 reference images into multi-shot video. The user sets a target duration and aspect ratio. The system plans the screenplay, dialogue, shots, and pacing, keeps characters and locations consistent, and assembles longer videos from multiple generated shots, up to ten minutes. The release does not state a resolution. VideoGen says it expects costs more than 50% below comparable models, based on internal validation, and that final pricing and third-party tests are still underway. In an internal experiment, an eight-step adapter cut average clip-render time by 58.2% against the same system's 28-step baseline. VideoGen says that result is an experimental configuration, not a guaranteed gain, and that a technical whitepaper is planned.
Outputs carry invisible watermarks identifying them as synthetic, and guardrails are meant to block adult and illegal content. A limited private-partner rollout began on 20 October, with general availability expected after more safety testing. A separate GlobeNewswire release on 8 September said the company had passed 5 million users and raised a $3.3 million seed round. Those figures are also the company's.
The credit pool
VideoGen is offering a $1 million pool of free credits for selected projects by creators, filmmakers, and studios, including students and amateurs. Applications opened on 20 October. The company plans to distribute credits through the end of 2026, and selection weighs the creative work and a viable distribution plan. Read the terms on rights to the work, and to any material you upload, before applying.
Claims with one source
- Every figure above comes from VideoGen. The release does not say whether "up to ten minutes" is one coherent pass or stitched clips. It does say longer videos are assembled from multiple generated shots.
- "More than 50% cheaper" has no published price behind it.
- A watermark that is invisible helps provenance only if a platform can read it.
This is the second ten-minute claim in a week, after Higgsfield's on 13 October. The companion piece today is a checklist for comparing them. For shots you generate on your own stack, RenderBob lets you choose local or cloud for each clip and keep the references with the studio.
What to watch
The whitepaper, a published price and resolution, the general-availability date, and how the credit programme chooses projects.
RenderBob 3 and this story
RenderBob 3's per-shot model choice is how you test a claim like this. Try a new model on one shot once it is available, and keep the rest of the cut where it was.
More from the blog
- Hugging Face's CEO Says Netflix's AI Documentary Went "Full Doomer Mode": What He Disputes and What the Sources Don't Settle
Clément Delangue, who appears in Netflix's Instadocs: AI Gone Wild, has criticised how the film tells the July attack on his company. The reports carry his side, and no reply from Netflix.
- Eleven Realtime: ElevenLabs' Speech-to-Speech Models Reach 90 Languages, With Two Sets of Latency Numbers
ElevenLabs announced Eleven Realtime on 20 October, two voice-agent models called Mini and Max. The latency figures differ between the two reports we read.