VideoGen unveils Model 1, claiming 10-minute multi-shot video from a single prompt
lmoroney · x · 2026-10-20
San Francisco startup VideoGen (YC S24) announced its first in-house video model, Model 1, which it says plans and generates up to 10 minutes of multi-shot video from one prompt while keeping characters and story consistent across scenes.
Key points:
- CEO Anton Koenig claims quality at 80% of ByteDance's Seedance 2.5 (native limit 30s) at less than half the cost; no benchmarks back this yet
- Official pricing and third-party validation are pending; AI Affairs could not verify claims
- Limited rollout to private partners began Oct 19, alongside a $1M pool of free AI credits
If the consistency claims hold, duration would be a notable leap for video generation, but all figures are company statements.
More from Multimodal
- Z-Image Turbo runs locally on Mac Studio: 8-step CFG-1 recipe yields striking B&W portraits — braddown · 2026-10-20
- Concatenative neural speech synthesis project shared — kastnerkyle · 2026-10-20
- MOSS Image with 8-step Turbo LoRA tested: 10x faster with minimal quality loss — Aggressive_Collar135 · 2026-10-20
- User turns their AI muse Maki into a punky ska song with Suno V6 — thederbiedone · 2026-10-20
- Grok Imagine parody short 'Wendy' shows off AI-made TV skits — Old_Put1306 · 2026-10-20
- AI-generated "Back to the Future" sequel clip goes viral for Halloween — mitchdeg · 2026-10-20