Flux 3 Image: Targeted Editing and 4K Output

According to Black Forest Labs, the new Flux 3 Image model introduces a multi-step editing feature designed to modify specific areas of a generated visual while keeping the rest of the composition completely intact. This update addresses a major workflow bottleneck for designers and creators who previously had to regenerate entire outputs to fix localized flaws.

How does multi-step editing change visual workflows?

Instead of relying on trial-and-error prompts to fix a single detail, creators can now isolate modifications to precise sections of the canvas. The model also allows users to compose complex scenes using bounding boxes and up to ten reference images simultaneously, removing previous limitations on style consistency across complex assets.

Outputs reach up to 4K resolution, supporting standard tasks like text-to-image, image-to-image, and improved text rendering. For teams integrating generation into production pipelines, API access is available at a 50 percent discount through October 8, while commercial weights allow local infrastructure hosting and fine-tuning. An open-weight version is scheduled for release in the coming weeks.