FLUX 3 Max

FLUX 3 Max is a video generation model developed by fal Research in collaboration with Black Forest Labs, launched on October 19, 2026. Built as a post-trained variant of the FLUX 3 foundation architecture, it is designed for text-to-video and image-to-video tasks featuring native synchronized audio. The model combines improved visual fidelity and enhanced prompt adherence with significant inference optimizations to accelerate generation pipelines.

Optimized for high-throughput workflows, FLUX 3 Max can produce a 10-second video clip in approximately 12 seconds while supporting variable durations between 5 and 20 seconds. It accommodates optional start and end image inputs for precise compositional control. The architecture emphasizes low latency without sacrificing output quality, making it practical for rapid creative iteration.