← Back to tools
F
Design

FLUX

Frontier AI image generation and editing models for creators and developers

Black Forest Labs' image generation model. State-of-the-art quality and prompt adherence.

01

About FLUX

FLUX is a suite of cutting-edge visual foundation models focused on high-quality image generation and editing. Developed by Black Forest Labs, it spans hosted production models such as FLUX.2 and open-weight models in the FLUX.1 family (including FLUX.1 [pro], [dev], [schnell], and the FLUX.1 Tools line for structured control). The system is designed to turn text prompts into detailed images and to manipulate existing images with inpainting, outpainting, and structural guidance, making it useful for both creative professionals and developers building visual applications.

Under the hood, FLUX.1 models use a rectified-flow transformer architecture trained with latent adversarial diffusion distillation to generate images in as few as 1–4 steps (for FLUX.1 [schnell]) while maintaining quality. Open-weight variants are distributed via Hugging Face and an official GitHub inference repo, enabling self-hosted or local GPU workflows. On the hosted side, the Black Forest Labs API exposes FLUX.2, FLUX Kontext, and other production-grade variants, with features such as 4 MP photorealistic outputs and multi-reference control. The FLUX.1 Tools set—Fill, Depth, Canny, and Redux—adds targeted capabilities for inpainting/outpainting, depth-map guidance, edge-map guidance, and image remixing/variation. These tools are available as models in the FLUX.1 [dev] series and are supported in ecosystems like ComfyUI.

FLUX is aimed at artists, designers, marketing teams, and video or VFX professionals who need high-quality still imagery as part of storyboards, concept art, key art, or asset creation pipelines, as well as engineers integrating image generation into products via API. It excels at fast, high-fidelity image synthesis from prompts, iterative refinement of existing images, and controlled transformations guided by depth or edge maps. Notable limitations include a focus on images rather than video or 3D, reliance on sufficient GPU resources for self-hosting, and adherence to usage and safety policies on the hosted API. For video professionals, FLUX is best thought of as a powerful still-image engine and asset generator that can plug into broader production and post-production workflows, rather than a direct text-to-video solution.

02

Key features

  • High-quality text-to-image generation with FLUX.1 and FLUX.2 models
  • Fast rectified-flow transformer architecture (1–4 step generation with FLUX.1 [schnell])
  • Advanced inpainting and outpainting via FLUX.1 Fill
  • Depth-map and canny-edge guided image generation with FLUX.1 Depth and FLUX.1 Canny
  • Image remixing and variation through FLUX.1 Redux adapters
  • Hosted API access to FLUX.2, FLUX Kontext, and FLUX.1 [pro]
  • Open-weight checkpoints for self-hosted and local GPU inference
  • ComfyUI workflows and nodes for Fill, Redux, Depth, and Canny models
  • Multi-reference control and 4 MP photorealistic output on FLUX.2
  • Apache-2.0 licensed open models suitable for commercial use (for FLUX.1 [schnell] and other open-weight releases)
03

Use cases

  • Generate high-quality concept art, key art, and style frames from text prompts for film, TV, and advertising.
  • Create variations and refinements of existing stills, plates, or key visuals using FLUX.1 Redux.
  • Perform precise inpainting and outpainting to remove, replace, or extend elements in production stills or marketing assets with FLUX.1 Fill.
  • Use depth or edge maps from 3D renders or on-set plates to guide stylized or photoreal re-interpretations via FLUX.1 Depth and FLUX.1 Canny.
  • Integrate image generation into pipelines and products (e.g., asset generators, design tools, or creative assistants) via the Black Forest Labs API.
  • Prototype look development and mood boards rapidly by self-hosting FLUX.1 models in local or on-prem environments.
04

Capabilities

API accessSelf-hostableWorks offlineCommercial useComfyUI nodeText → Image
05

Tech specs

Max resolution
4MP (hosted FLUX.2); varies for open models
Base model
FLUX.1 rectified-flow transformer family; FLUX.2 production image models
06

Platforms & integrations

Platforms
WebWindowsmacOSLinux
Integrates with
ComfyUIHugging FaceReplicateFal.aiDiffusers
Output
Image
Workflows
Pre-productionPost
Tagged
text-to-imageimage-editinginpaintingdepth-guidanceedge-guidanceopen-weightsAPI