Is FLUX 3 available to the public yet?

Not fully. FLUX 3 Video is in early access for approved partners, FLUX 3 Image is due to open “in the coming weeks,” and an open-weight FLUX 3 Dev release is planned for later in 2026.

How is FLUX 3 different from earlier FLUX models?

Earlier FLUX releases generated still images only. FLUX 3 is trained jointly on images, video, and audio in one model, and its video-prediction approach also powers a robotics component called FLUX 3 Action.

Does FLUX 3 work with robots?

Yes. Built with robotics startup mimic robotics, FLUX 3 Action predicts physical motion, and Black Forest Labs says it can be fine-tuned for a new task with about 30 minutes of robot data; Audi is testing it for production-line work.

How much does FLUX 3 cost?

Black Forest Labs has not disclosed pricing. Access is currently limited to an early-access application; API and paid access are planned as the rollout continues.

Are Black Forest Labs' benchmark claims independently verified?

No. The head-to-head preference results — such as FLUX 3 beating Runway Gen-4.5 in 77% of comparisons — come from the company’s own testing, and the methodology has not been published or independently reviewed.

Black Forest Labs Launches FLUX 3, a Multimodal AI Model

Black Forest Labs, the German AI lab behind the FLUX image models, launched FLUX 3 in early access on July 23 — its first system trained to generate images, video, and audio within a single model rather than stitching together separate tools.

What’s new

Previous FLUX releases handled still images only. FLUX 3 Video generates clips up to 20 seconds long with synchronized audio — dialogue, sound effects, and ambient noise matched to the visuals — from text prompts, existing images, or other video, and supports multilingual dialogue and extending a clip from a fixed keyframe. A companion image-generation mode, FLUX 3 Image, is due to open to early access “in the coming weeks,” the company said, and an open-weight FLUX 3 Dev release is planned for later this year alongside API access.

Black Forest Labs says internal human-preference testing found reviewers favored FLUX 3 video clips over rival models Runway Gen-4.5 in 77% of comparisons and Luma Ray 3.2 in 93%. The company has not published its testing methodology, and the figures have not been independently verified.

A push into robotics

FLUX 3’s video backbone doubles as a foundation for robotics: a new component called FLUX 3 Action, built with robotics startup mimic robotics, predicts physical motion the same way the model predicts video frames, treating a robot arm’s next move like the next frame in a clip. Black Forest Labs said the approach can be fine-tuned for a new manipulation task with as little as 30 minutes of robot data, and that automaker Audi is testing it for production-line work — a use case similar in spirit to the action-prediction approach behind Vision-Language-Action models already deployed in humanoid robots.

Founded in 2024 by former Stability AI researchers Robin Rombach, Andreas Blattmann, and Patrick Esser, Black Forest Labs raised $300 million in a Series B round in December 2025 at a $3.25 billion valuation. Pricing for FLUX 3 has not been disclosed; access is currently limited to companies that apply for early access.

What’s new

A push into robotics

Frequently asked questions