Black Forest Labs releases FLUX 3: up to 20 seconds of video with sound from a single model

On July 26, 2026, Black Forest Labs released FLUX 3, a multimodal flow model that generates images, video up to 20 seconds long with native sound, and audio, and predicts robot actions, all from a single set of weights. It is the first model in the FLUX family in which video, sound and action prediction come from a shared architecture.

What FLUX 3 Video can do

FLUX 3 Video produces clips up to 20 seconds long in a single generation, with sound included. According to Black Forest Labs, the model supports five modes: text-to-video, image-to-video, reference-based video-to-video, keyframe-to-video for controlled transitions, and video-audio continuation from an input clip.

The company also highlights multilingual dialogue, stitching several clips into one multi-scene sequence, and animated typography generation. The BFL team claims strong handling of human facial expressions and accurate alignment of sound with physical events: an impact on screen matches the sound of the impact.

• Released July 26, 2026
• Video up to 20 seconds with native sound in a single generation
• Five modes: text/image/video/keyframe-to-video and video-audio continuation
• FLUX-mimic: the robot policy runs in under 80 ms on a single RTX 5090
• Access is staged: Video and Action are in early access, Image comes later, open weights come last

How the model works

FLUX 3 is built on Self-Flow, a Black Forest Labs method that combines flow matching with self-supervised feature reconstruction in a single architecture. The company introduced Self-Flow itself back in March 2026. What is new here is the scale: according to BFL, compute and data volume were "significantly increased," and the model was trained on video, images and sound at the same time.

Resources are split unevenly. According to Black Forest Labs, video prediction takes more than 95% of training compute, while audio accounts for less than 0.5% of tokens. The idea is that the modalities constrain each other during training.

"No single modality gives a complete description of the world," is how the

Black Forest Labs research team states its principle.

How far ahead of competitors is FLUX 3?

In a preliminary human-preference test, FLUX 3 beat most of its rivals. The comparison used 10-second text-to-video clips in 720p with sound.

• Luma Ray 3.2: FLUX 3 was preferred in 93% of comparisons
• Runway Gen-4.5: 77%
• Grok Imagine Video: up to 69%; Kling v3 Pro: 60%
• Seedance 2.0 and Gemini Omni Flash: 52%, nearly a coin flip

The same backbone drives FLUX-mimic, a robot policy that, according to BFL, runs in under 80 ms on a single RTX 5090 GPU. This shows that the multimodal model applies not only to media generation but also to controlling physical actions.

What this means

FLUX 3 is a bid for a single foundation model in which image generation, video with sound, and robot control grow out of shared weights. Access is still limited: Video and Action are in early access, Image will follow later, and Black Forest Labs says it will release open weights last.

Frequently asked questions

What is FLUX 3?

FLUX 3 is Black Forest Labs' multimodal flow model, trained simultaneously on images, video and sound. From a single set of weights, it generates images and video up to 20 seconds long with native audio, and predicts robot actions.

When will FLUX 3 open weights be released?

There is no exact date. Access is staged: Video and Action come first in early access, then Image, and Black Forest Labs plans to publish the open weights last.

What does a FLUX 3-based robot run on?

FLUX-mimic, a robot policy built on the same backbone. According to Black Forest Labs, it responds in under 80 ms on a single NVIDIA RTX 5090 GPU.