# FLUX 3 Max: fal and Black Forest Labs release a video model that makes 10 seconds in 12

> On October 19, 2026, fal and Black Forest Labs released FLUX 3 Max, a post-trained version of the FLUX 3 video model. According to fal, it renders a 10-second 720p clip in about 11–12 seconds, 22× faster than Seedance 2.5, and it ties for first on an independent image-to-video leaderboard. The speed claims haven't been independently verified yet.

Published 2026-10-24 · Updated 2026-10-24
URL: https://ai2face.com/news/flux-3-max

On October 19, 2026, fal and Black Forest Labs released **[FLUX 3](/models/flux-3) Max**, a post-trained version of the FLUX 3 video model. BFL provided the base model, and fal's research team post-trained it and optimized inference. BFL itself describes FLUX 3 as a model designed for this kind of customization and features Max on its FLUX 3 Video page.

## What FLUX 3 Max can do

- **Text-to-video and image-to-video.** These are the two main modes at launch. fal's press release also mentions a keyframes mode.
- **720p, 5 to 20 seconds, with sound.** According to Artificial Analysis, the model generates a clip together with its audio track, like the original [FLUX](/models/flux) 3 Video.
- **Commercial use.** fal allows commercial use of videos generated through its API, under its terms.

## How fast it is

Speed is the headline. In fal's tests, FLUX 3 Max renders a 10-second video in 10.95 seconds (per the press release) or 12.45 seconds (per the model page). In the same test:

| Model | 10 seconds of video | Image-to-video rating |
|---|---|---|
| FLUX 3 Max | 10.95–12.45 s | 1,139 |
| FLUX 3 | 126.44 s | 1,089 |
| [Seedance](/models/seedance) 2.5 | 277.8 s | 1,139 |

That's where the "22× faster than [Seedance 2.5](/models/seedance-2-5) at the same quality" claim comes from. One caveat: the speed figures are fal's own, and the company hasn't published a methodology or independent tests.

## Where it ranks

Quality is independently confirmed. On the Artificial Analysis leaderboard, FLUX 3 Max ranks second in image-to-video — statistically tied for first — and fourth in text-to-video. Its rating of 1,139 from 4,868 votes matches Seedance 2.5 and is 50 points above the original FLUX 3. The ranking is still preliminary.

## Pricing

On fal, a second of 720p video costs $0.17, with no subscription or minimum. Until November 7, 2026, a 75% discount brings it to $0.0425 per second — about $0.43 for a 10-second clip.

## How to try it in AI2Face

FLUX 3 Video and FLUX 3 Max are available in the AI2Face generator. Next to them are other strong video models: Seedance 2.5 with clips up to 30 seconds, [Gemini Omni](/models/gemini-omni) and [Hailuo](/models/hailuo). Run one prompt on several models and keep the best result. The cost is shown before you start. All the options are in our guide to [AI video generators](/blog/nejroseti-dlya-video).

## FAQ

### What is FLUX 3 Max?

A post-trained version of Black Forest Labs' FLUX 3 video model. fal's research team built it with BFL: BFL provided the base model, and fal improved quality and sped up generation.

### How much faster is FLUX 3 Max?

In fal's tests, it renders a 10-second clip in 10.95–12.45 seconds. Seedance 2.5 takes 277.8 seconds in the same test, and the original FLUX 3 takes 126.44 seconds. These are vendor numbers; there's no independent check yet.

### What resolution and length does FLUX 3 Max support?

720p and clips of 5 to 20 seconds with sound, according to Artificial Analysis. Modes are text-to-video and image-to-video.

### How much does FLUX 3 Max cost?

On fal, $0.17 per second of 720p video. Until November 7, 2026, a 75% discount brings it to $0.0425 per second.

## Sources

- https://fal.ai/flux-3-max
- https://x.com/bfl_ai/status/2112255353677328415
- https://x.com/ArtificialAnlys/status/2112266414593827110
- https://bfl.ai/models/flux-3-video
- https://finance.yahoo.com/technology/ai/articles/fal-launches-flux-3-max-194600494.html