Wait Which Model?
← Back to Video models
Black Forest LabsNotable

FLUX 3

Released Jul 23, 2026

Black Forest Labs' multimodal foundation model, trained jointly on images, video and audio. This entry covers its video generation (FLUX 3 Video), which makes clips of up to 20 seconds with native audio from text, images, keyframes or an input clip.

Price per second
$0.17
Max clip length
20 s
Max resolution
4K
Native audio
Yes
AA arena Elo
1142
LMArena Elo
1493
Capabilities
text to video, image to video, extension
Licence
Proprietary
Strengths & weaknesses

Strengths

  • Clips up to 20 s with native audio, including multilingual dialogue and lip-sync
  • Outputs up to 4K (uhd, 3840x2176 at 16:9) at 24 fps
  • Keyframes, multi-shot sequences and video continuation
  • Cheap draft mode ($0.06/s) that can be enhanced to full quality

Weaknesses

  • BFL docs still label FLUX 3 a preview model
  • No open weights yet (FLUX 3 Dev is announced but unreleased)

FLUX 3 is a single multimodal family (video, image, action). This entry is its video generation only, and FLUX 3 Image is separate. releaseDate is the Jul 23, 2026 announcement, which opened FLUX 3 Video in Early Access (AA lists 'Jul 2026'). General availability via the BFL API and partners was announced Aug 4, 2026 (bfl.ai/blog/flux-3-video). The docs still say 'FLUX 3 is a preview model'. Audio is on by default and can be disabled. Duration is 5 to 20 s or auto. The blog's own internal Elo chart lists FLUX 3 T2V at 1135, which is not used. AA silent board: 1091 (#5). LMArena score is the preliminary 'flux-3-video' row (#3) from the board dated Sep 21, 2026. The blog mentions video-to-video generation from a reference clip, and editing ships as a separate product (FLUX Video Edit [fast], /v1/flux-tools/video-edit-v1, $0.03/s), so 'video-editing' is not claimed. fal ids come from fal's own model pages. fal also has first-last-frame-to-video, keyframes-to-video, /draft twins and draft-enhance endpoints.

Links

API model strings

  • bflflux-3-video
  • falblackforestlabs/flux-3/text-to-video
  • falblackforestlabs/flux-3/image-to-video
  • falblackforestlabs/flux-3/extend-video