fal launches FLUX 3 Max video model, 22 times faster than Seedance 2.5 at the same reported quality

fal launched FLUX 3 Max on October 19, 2026, a video generation model post-trained by fal Research with Black Forest Labs on the FLUX 3 foundation model. fal reports a 1,139 Elo rating in preliminary Artificial Analysis Video Arena human preference evaluations, a 50 point gain over the original FLUX 3, and generation of a 10-second video in 10.95 seconds in its latency benchmark, which it describes as 22 times faster than Seedance 2.5 at the same reported Elo and 11.5 times faster than the original FLUX 3. It is available through fal's inference platform for text to video, image to video and keyframes to video, at 75% off for the first three weeks.

The figures are fal's reported results from its own release and benchmarks; the Elo rating is described as preliminary. No physics, shot sequencing or dialogue capability jump is claimed in the release: the launch is a speed and cost story.

Sources: PR Newswire; Unite.AI