Fal Debuts FLUX 3 Max Video Model Post-Trained With Black Forest Labs

Summary by forck

fal announced 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, available the same day through fal's inference platform. fal Research continued training the FLUX 3 model to improve visual fidelity, motion coherence, prompt adherence and generation performance, and the product page describes post-training, inference optimization and systems engineering behind it. fal reported a 1,139 Elo rating across 4,868 votes in preliminary independent human preference evaluations on Artificial Analysis Video Arena, a 50-point gain over the original FLUX 3's 1,089.

• Its latency benchmark measured a 10-second video at 10.95 seconds, which fal described as 22 times faster than Seedance 2.5's 277.80 seconds at the same reported Elo rating and 11.5 times faster than the original FLUX 3's 126.44 seconds.
• The product page publishes a different figure, a 12.45-second average for a 10-second 720p video, and describes the improvement over FLUX 3 as roughly 10 times faster.
• Artificial Analysis's image-to-video leaderboard listed FLUX 3 Max at 720p with 1,138 Elo plus or minus 15 across 5,054 samples, ranked second behind MiniMax H3 at 768p with 1,140 Elo.