Black Forest Labs has released Flux 3 Image, enabling multi-step edits that only modify targeted areas while preserving the rest of the image. The model supports up to 4K output and composition tools with multiple reference images.
Black Forest Labs' new Flux 3 Image model adds focused editing capabilities to the company's generative AI toolkit. The key feature allows users to perform multiple sequential edits without affecting unmodified portions of an image—a significant improvement over earlier tools that often required regenerating entire pictures.
Composition and Control
The model includes bounding box controls for precise scene composition, letting users define exact areas for generation or modification. Users can reference up to ten images simultaneously, providing more granular control over the creative process.
Resolution and Quality
Flux 3 Image supports output resolutions up to 4K, addressing professional use cases that demand higher fidelity. This positions the model for applications beyond social media graphics and quick mockups.
Open Weights Coming
Black Forest Labs plans to release open weights in the coming weeks, following the company's pattern of making models accessible to developers and researchers. This move typically enables faster adoption and community-driven improvements.
Context
Flux 3 Image represents the image generation component of Black Forest Labs' broader Flux 3 model family. The company has positioned itself as a competitor to established players like OpenAI's DALL-E and Midjourney by emphasizing speed, quality, and accessibility.
The multi-step editing capability addresses a long-standing friction point in AI image generation: the need to regenerate entire images for minor tweaks. Black Forest Labs' approach allows iterative refinement without disrupting completed portions of artwork.
Open weight availability typically follows commercial release by weeks, giving the company time to refine performance and gather user feedback before broader distribution.
Experts warn that artificial intelligence could amplify disinformation campaigns during elections across Southeast Asia, a region with young, digitally connected populations vulnerable to manipulation.
DeepSeek has released Harness Desktop, bringing its AI capabilities to macOS and Windows users. The application provides local access to DeepSeek's models across both platforms.
According to the latest Ramp AI Index, US companies are increasing their AI adoption while simultaneously reducing spending on these technologies. The trend suggests improving cost efficiency in enterprise AI implementation.
Cloudflare has released Clef and Clef-flash, open-weight multimodal decision models built on Qwen foundations. The company claims both models outperform Jev in speed and capability.