Black Forest Labs has launched FLUX 3 Image, a new AI model that turns image editing into a controlled, box-by-box process. The company says the core idea is simple: every element you place on the canvas stays where you put it, and anything you did not touch remains exactly as it was.
The model covers both generating images from scratch and editing a finished picture one bounding box at a time. Black Forest Labs is also offering it through an API and as open weights, with commercial licensing available, which makes FLUX 3 Image relevant to developers and businesses as well as individual creators.
FLUX 3 Image takes the guesswork out of AI editing
Most AI image tools regenerate large parts of a picture when you ask for a change. Black Forest Labs is pitching FLUX 3 Image as a more precise alternative, where edits are confined to the area you select and the untouched parts of the image are preserved.
The workflow is built around bounding boxes. You lay out an image exactly how you want by placing elements in defined areas, then edit those areas individually until the result matches your intent.
FLUX 3 Image specs and availability
Black Forest Labs lists FLUX 3 Image across its usual channels: models, research material, an API, open weights, pricing and enterprise options. Developers can read the documentation, and companies can contact the team about commercial weights.
The company has not published a full technical specification sheet or a release date for broader consumer products in the material available, so those details remain to be confirmed.
- Compose images from scratch on a canvas
- Edit a finished image one bounding box at a time
