FLUX 3 Max: fal and Black Forest Labs launch video model that generates 10 seconds in about 12 seconds

fal announced on October 19, 2026 the FLUX 3 Max, a video generation model post-trained in partnership with Black Forest Labs. The model is available on the fal platform for text-to-video, image-to-video, and keyframes-to-video generation.

What happened?

According to the official announcement, FLUX 3 Max was developed from the FLUX 3 foundational model by Black Forest Labs. fal Research applied specialized post-training, reinforcement learning, and inference optimizations. The result, according to data released by fal, places the model among the highest quality in independent image-to-video evaluations and with significantly lower latency.

In fal's own benchmarks, a 10-second video is generated in about 10.95 seconds. The company claims this represents more than 22 times the speed of Seedance 2.5 at the same reported Elo level and 11.5 times faster than the original FLUX 3 (which took 126.44 seconds in the same test).

Why it matters

AI video generation still faces the trade-off between visual quality and processing time. Faster models make iteration easier for creators, game developers, content producers, and teams that need quick previews. fal highlights that post-training and systems optimization can raise quality and speed at the same time, rather than forcing a choice between the two.

Robin Rombach, co-founder and CEO of Black Forest Labs, commented that FLUX 3 was designed as a foundation for others to build on, and that Max is an example of that collaboration. Burkay Gur, CEO of fal, said important advances will come from jointly optimizing models and inference systems.

In practice

The model has been available since October 19 on the fal platform. For the first three weeks (until around November 7, according to fal information), there is a 75% discount. Endpoints include text-to-video and image-to-video, with keyframe support.

The Elo numbers (1,139 in preliminary evaluations on Artificial Analysis Video Arena) and latency figures are those reported by fal and by independent evaluations cited in the announcement. As with any model launch, real-world results may vary depending on prompt, resolution, and specific use.

Sources

Transparency: This content was created, edited or reviewed with the aid of artificial intelligence. Information was cross-checked with public posts on X and sources available on the internet. Check the original sources for the full context.

By GeekikiBot