FLUX 3 Image arrives with bounding-box layouts and pixel-exact edits

Black Forest Labs has launched FLUX 3 Image, the image model of its FLUX 3 family, through its API. It lets users place elements with bounding boxes, keeps untouched pixels exactly the same during edits, combines up to ten references and outputs up to 4K. The company says an open-weights version is

Paylaş
FLUX 3 Image arrives with bounding-box layouts and pixel-exact edits

Black Forest Labs launched FLUX 3 Image, its image generation and editing model, on October 1, 2026. The model is the image part of the FLUX 3 family the company unveiled on July 23. Its focus is control: every element in an image can be placed with a bounding box, and pixels that an edit does not touch stay exactly the same.

The image model of the FLUX 3 family

According to the model page, FLUX 3 is Black Forest Labs' multimodal model for video, audio, images and actions, and FLUX 3 Image is the part that generates and edits images. FLUX 3 Action, an open model from the same family that plans robot movements, was released on September 23. FLUX 3 Image runs through a single API endpoint, and according to the release notes, the prompt itself decides whether a request generates a new image or edits one.

Bounding-box layouts and pixel-exact edits

According to the release notes, users can add a JSON list to the prompt to place every element of a scene with a bounding box. According to the documentation, boxes are defined on a 0 to 1000 grid regardless of the image's aspect ratio. For editing, the model handles local operations such as recoloring, replacing, moving, resizing and removing elements. Regions to keep are marked in the prompt as "anchor"; the company describes these edits as pixel-exact and says the other pixels stay unchanged across multi-turn edits.

The model can combine up to ten reference images in a single request, referred to by their position in the prompt. Output ranges from a small size suited to fast previews up to 4K, about 16 megapixels, in 15 aspect ratios.

Access and licensing

FLUX 3 Image is available through the Black Forest Labs API and the company's web-based Playground. According to the model page, companies running image generation at scale can license commercial weights to fine-tune and deploy the model on their own infrastructure. According to the company's post on X, an open-weights version of FLUX 3 Image is launching in the coming weeks; no exact date was given. Current prices are on the company's site.

The examples the company shared are images it chose itself. To see how pixel preservation holds up in complex scenes and long chains of edits, it is worth testing the model on your own work. For a fuller picture, follow evaluations from independent researchers as well.

Sources: Black Forest Labs: FLUX 3 Image, Black Forest Labs: release notes, Black Forest Labs: FLUX 3 Image documentation, Black Forest Labs: announcement on X.

Devamını oku

Mistral تطلق نموذج Voxtral مفتوحا لتحويل الكلام العربي إلى نص فوريا

Mistral تطلق نموذج Voxtral مفتوحا لتحويل الكلام العربي إلى نص فوريا

أطلقت Mistral AI نموذج Voxtral Mini 4B Realtime Arabic الذي يحول الكلام بالعربية ولهجاتها إلى نص في الوقت الفعلي، بترخيص Apache 2.0. وإلى جانبه نموذج LIDstral Arabic الذي يحدد لهجة النص المكتوب بالحروف العربية أو لغته. وقد طُوّر النموذجان بالتعاون مع وزارة الانتقال الرقمي المغربية.

GlobalFeed Editor tarafından