fal and Black Forest Labs bring video generation close to real time with FLUX 3 Max

What is changing?

fal has launched FLUX 3 Max, a video model developed with Black Forest Labs. It adapts the FLUX 3 foundation through specialized post-training, reinforcement learning, and inference optimization. Text-to-video, image-to-video, and keyframe-to-video generation are available through the same API.

In fal’s published measurements, the model generated a 10-second clip in 10.95 seconds; Black Forest Labs summarizes the result as roughly 12 seconds on its product page. The provider also reports a preliminary human-preference score of 1,139 Elo from 4,868 votes on Artificial Analysis Video Arena and 11.5× faster inference than the original FLUX 3. These should be treated as provider-reported results that depend on the model and infrastructure conditions.

FLUX 3 Max is available on fal’s platform from October 19. A 75% introductory discount applies for the first three weeks. Faster generation could let advertising and short-video teams test more drafts in the same amount of time, but quality, cost, and latency should still be measured in real production workflows.