Flux 3

• - 🏢 From Black Forest Labs, the German lab behind the Flux family.
• - 🆕 Editing variant of Flux 3, per this catalog listing.
• - 🔧 Targets inpainting, object changes and broader scene edits.
• - 👁️ Works from an input image plus natural-language edit instructions.
• - 📚 Successor in the Flux editing lineage after Flux 2 Max Edit.
• - 🎯 Catalog description cites photorealistic fidelity as its focus.
• - 🌐 Sits alongside Flux 3 text-to-image and video siblings.
• - 🔒 No public weights or technical report surfaced for this variant.

Black Forest Labs

Black Forest Labs is a generative AI company based in Freiburg im Breisgau, Germany, founded by former members of Stability AI. The lab is best known for developing the Flux family of text-to-image models, which generate images from natural language prompts…

Flux 3 (image edit) is the editing-focused member of the Flux 3 generation, per this catalog listing. Black Forest Labs publishes a family of Flux models spanning text-to-image and image-to-image work, several of which support editing existing images rather than only generating from scratch. This variant is used by supplying a source image and an instruction or mask, and it returns an edited result; the catalog description characterises it as targeting photorealistic fidelity across inpainting, object replacement, and wider scene changes.

Within the family, and again per this catalog listing, it pairs directly with Flux 3, the text-to-image model listed with the same release date, and succeeds Flux 2 Max in the editing lineage. The Flux 2 generation it follows includes Flux 2 Pro and Flux 2 Max for generation and editing.

Flux 3 also extends beyond stills: the catalog lists in the same generation Flux 3 text-to-video, Flux 3 image-to-video, and Flux 3 First Last Frame, so a still edited here can be carried into motion work.

Black Forest Labs has not published a public technical report or model card detailing Flux 3's editing architecture, parameter count, or benchmark results, so no performance figures are quoted here. The earlier Flux.1 Kontext line remains documented by the lab for instruction-based editing, with open weights released for its development variant.