88Z AI, 24 July — Black Forest Labs has released FLUX 3, a multimodal foundation model that jointly learns images, video, and audio within a unified architecture. Built on Self-Flow technology, it supports text-to-video and image-to-video generation, producing videos with native audio up to 20 seconds long in a single pass, while demonstrating strong performance in robotic manipulation tasks.
At a glance
- FLUX 3 jointly learns images, video, and audio in a unified architecture, supporting single-pass generation of 20-second videos with native audio.
- Built on Self-Flow technology, the model learns world representations via physical constraints, excelling in robotic manipulation tasks.

The update
Black Forest Labs has officially launched FLUX 3, a multimodal foundation model. This marks the first time that images, video, and audio are jointly learned within a unified architecture. At its core, the model utilizes Self-Flow technology to efficiently handle various generation tasks, including text-to-video and image-to-video modes. In terms of generation capabilities, FLUX 3 supports the single-pass creation of video clips up to 20 seconds long, complete with synchronized native audio, eliminating the need for post-production dubbing. Furthermore, by learning world representations through physical constraints across modalities, the model gains a better understanding of physical laws, resulting in superior performance in tasks requiring precise physical interaction, such as robotic manipulation.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.