FLUX Video
by Black Forest Labs · Model · Video
promising
Near the top of Arena video rankings, but the promised open weights have not appeared
Unified multimodal model for image, video with audio (up to 20 s) and robot action prediction; preview with early-access signup only
Why we say so
Arena ranks FLUX 3 Video second on text-to-video in both September (1493) and November (1491). It sits well behind Gemini Omni 1.2 Flash (1540) and just ahead of Grok Imagine Video 1.5 (1488). AI Search reported in October that the open-weights promise made months earlier still had no release.
Ranks #2 for video generation
Evidence
- Nov 9#2 on Arena text-to-video (1491)
- Oct 4Open-weights promise made months ago, still no sign of a release
- Sep 22#2 on Arena text-to-video (1493)
- Jul 26Self-reported preference tests rank 10 s 720p clips slightly above Gemini Omni Flash and Seedance 2.0; reviewer doubts it matches Seedance quality
- Jul 26Expected to follow earlier pattern: paid closed max model via API plus a lower-quality open-weights dev version