Microsoft Launches In-House AI Models to Cut Costs
Microsoft has launched two new internally developed artificial intelligence models as it works to reduce the cost of operating AI …
Black Forest Labs Launches FLUX 3 for Images, Video and Audio
Black Forest Labs has introduced FLUX 3, a multimodal artificial intelligence model designed to generate and edit images, video and synchronized audio within one system.
Unlike earlier FLUX releases focused mainly on image generation, the new model was trained across images, video and audio at the same time. The company says this unified approach helps FLUX 3 understand movement, physical interactions and how sounds relate to visual events.
Video with native audio
FLUX 3 can generate videos with native audio lasting up to 20 seconds in a single request.
The model supports text-to-video, image-to-video and video-to-video generation. Users can provide a starting image, visual reference or existing clip and ask the system to create a new scene while preserving important elements such as characters or style.
It can also continue existing video and audio, create transitions between selected keyframes and produce multilingual dialogue.
Black Forest Labs says individual clips can be connected into longer multi-scene sequences. Visual references are intended to help maintain character consistency across those scenes.
Image generation and editing
FLUX 3 also supports image creation and editing across different resolutions, styles and aspect ratios.
The company says the model has improved its ability to follow complex prompts and render accurate text in multiple languages compared with earlier FLUX systems.
However, the image version is not yet broadly available. Black Forest Labs plans to begin its early-access rollout in the coming weeks.
Early benchmark claims
Black Forest Labs published preliminary preference tests comparing FLUX 3 with several competing video models.
The company said testers preferred its output over Runway Gen-4.5 in 77% of comparisons and over Luma Ray 3.2 in 93%. It also reported smaller advantages against Kling v3 Pro, Grok Imagine Video and several other systems.
These figures are based on early internal evaluations. Black Forest Labs said it will release fuller benchmark results and methodology when the model becomes more widely available.
Limited early access
FLUX 3 Video is currently available only through an early-access program.
Selected developers and companies can apply for access, while broader API availability and private model-weight access are planned for later stages. FLUX 3 Image will follow, along with faster and open-weight versions expected later in 2026.
Companies already testing the model include Canva, Krea, Picsart, Burda and Magnific.
From content creation to robotics
Black Forest Labs is also extending FLUX 3 beyond media generation.
The company has partnered with mimic robotics to create FLUX-mimic, a model that uses the FLUX 3 video foundation to predict robot actions. The system is being tested in manufacturing environments, including projects involving Audi.
This reflects the company’s wider strategy of using one multimodal model for creative tools, simulation and physical AI.
FLUX 3 therefore represents a major expansion of the FLUX product family. Its ability to combine images, video, audio and action prediction could make it useful across advertising, film production, design, e-commerce and robotics.
Its wider impact will depend on output quality, pricing, safety controls and how quickly Black Forest Labs moves beyond the current limited release.
Sign up to our newsletter
Receive our latest updates about our products & promotions
Related articles
SpaceX Launches Starlink V3 Satellites but Suffers Another Booster Failure
SpaceX successfully deployed its first operational Starlink V3 satellites during the 13th test flight of Starship, marking a major step …
Anthropic Launches Claude Opus 5 for Advanced AI Tasks
Anthropic has launched Claude Opus 5, its latest high-performance artificial intelligence model for coding, research, financial analysis and long-running agent workflows. …