FLUX 3 Max ties the top AI video models and is far faster, by fal's own timing
fal says FLUX 3 Max makes a 10-second clip in about 12 seconds, 22 times faster than Seedance 2.5. The speed is fal's own test, and the quality ranking is an early statistical tie, not a win.
fal and Black Forest Labs released FLUX 3 Max on Oct 19, a video model that makes clips from a text prompt or a still image. Black Forest Labs built the FLUX 3 model underneath it; fal did the extra training and the engineering that makes it run faster, and sells access on its own platform.
The big claim is speed. By fal's own timing, a 10-second clip at 720p comes back in about 12 seconds, where Seedance 2.5 takes close to five minutes. That is where the 22x comes from. fal's press release and its product page give different times (10.95 and 12.45 seconds), so treat the exact figure as the company's number, not a measured standard.
On quality, the outside check is Artificial Analysis's blind vote board, where people pick the better of two clips made from the same image. FLUX 3 Max sits second there, inside the margin of error with MiniMax H3 and Seedance 2.5. So "same quality" means a statistical tie in an early ranking, not a win, and nobody has put it through a real production yet.
Why it matters: when every take costs minutes, people stop trying new versions early. If a usable shot really comes back in seconds, the model you sketch with and the model you finish with can be the same one. The list price is $0.17 per second of video at 720p, or $1.70 for a 10-second clip, cut by 75% until Nov 7.
Source: fal press release and product page, Artificial Analysis leaderboard, Oct 19-20, 2026.








