FLUX 3 Max is live on fal for developers generating video from a text prompt, a still image, or keyframes. fal Research post-trained the model with Black Forest Labs on FLUX 3. In the comparison fal published with the launch, a 10-second clip takes 12.45 seconds at 1,139 image-to-video Elo, against 126.44 seconds and 1,089 Elo for the original FLUX 3 and 277.80 seconds at the same 1,139 Elo for Seedance 2.5. Preliminary Artificial Analysis evaluations cited in the release count 4,868 votes for that 1,139 Elo, with overlapping confidence intervals among the highest-ranked image-to-video models. fal Research did the additional post-training, reinforcement learning, and inference optimization on its own serving stack, working from the FLUX 3 base built by Black Forest Labs. That base model generates clips of up to 20 seconds with synchronized audio. On fal, FLUX 3 Max is available for text-to-video, image-to-video, and keyframes-to-video, at 75% off for the first three weeks after the October 19, 2026 launch. fal is a San Francisco generative media platform founded in 2021 by Burkay Gur and Gorkem Yurtseven. Developers use it to call more than 1,000 image, video, audio, and 3D models through one API, and to run custom models on serverless GPUs. FLUX 3 Max follows H3 Max, fal Research's earlier post-trained video model, and applies the same approach to a Black Forest Labs foundation model the platform already hosts.