Black Forest Labs launches FLUX 3 for image, video, and audio generation

Black Forest Labs, the Freiburg, Germany-based AI lab, unveiled FLUX 3 today, a multimodal frontier model that marks a significant expansion beyond its earlier image-focused FLUX lineup. FLUX 3 is trained to understand and generate images as well as combined audio and video clips up to 20 seconds, all from a single prompt. Critically, the company emphasizes that FLUX 3 is jointly trained across those modalities—image, video, and audio—rather than combining pre-built separate models under one interface. This architectural choice underpins Black Forest Labs' core pitch to enterprises: that creative generation, simulation, computer use, and robotics should be understood as connected applications of a single underlying capability. The company frames this as "visual intelligence," referring to models that can perceive, predict, and act across both physical and digital environments. The release also marks Black Forest Labs' first public video generation model. FLUX 3 will be offered through four product lines: FLUX 3 Video, FLUX 3 Image, and FLUX 3 Act. The initial release is limited, indicating the company is rolling out access in phases rather than an immediate, broad launch.

Black Forest Labs has positioned FLUX 3 as a departure from the modular approach that has dominated multimodal AI: rather than building independent image, video, and audio generators and bolting them together, the company trained a single architecture jointly across all three domains. This design choice reflects a broader strategic vision where the company sees image generation, video creation, audio synthesis, robotic perception, and robotic control as manifestations of a unified capability—what it calls "visual intelligence." The framing suggests that enterprises should think of these traditionally separate applications as connected rather than siloed, potentially opening new workflows where a single model handles perception and action across physical and digital contexts. The limited initial release indicates the company is managing rollout carefully, likely to gather feedback and validate performance before scaling.

No comments yet. Be the first to share your thoughts!