
FLUX.2 can use up to 10 reference images to keep characters, products, and styles consistent. Refer to specific images in your prompt (like “use the outfit from image 2 on the character in image 1”) and keep identity and styling stable across shots.

Compared to earlier models, FLUX.2 delivers richer textures, cleaner materials, and more stable lighting. It is ideal for product shots, interior visualizations, and photography-style scenes where realism really matters.

FLUX.2 follows complex, structured instructions, including JSON-style prompts and composition constraints. Describe different regions, elements, or panels, and you can even define exact hex colors such as #DDC57A for brand palettes or UI components.

Edit images up to 4 megapixels (for example, 1920×1920) while preserving details and coherence. FLUX.2 handles flexible aspect ratios, making it easy to create blog hero images, social banners, and product visuals that are ready to publish.

FLUX.2 includes [pro] and [flex] hosted variants, a powerful open-weight [dev] model, and the upcoming [klein] Apache 2.0 edition. This gives you a full range of options from fast hosted generation to deeply customizable local workflows.

Describe the scene, characters, layout, and styles you want. Optionally reference specific images by index, like “image 1” and “image 2”.
Upload images for editing, add reference photos for consistency, or start from text-only generation using FLUX.2 on GoEnhance.
Adjust steps and guidance, explore variations, then export your results in PNG/JPG/JPGE or upscale for high-resolution delivery.
FLUX.2 is now available through GoEnhance AI. Experience multi-reference editing, high-resolution workflows, and precise text rendering in your browser.