Flux

Free

Text-to-image models by Black Forest Labs with high-quality photorealistic output. #opensource

FreeFree tier

Outputs: image

Type

Open Source

Company

Black Forest Labs

Visit Website GitHub (40 stars)

About Flux

Flux is a suite of open-weight text-to-image models developed by Black Forest Labs. This official inference repository provides minimal code for image generation, editing, and structural conditioning tasks. Models include FLUX.1 [schnell] (Apache 2.0 licensed for fast generation), FLUX.1 [dev] (high-quality non-commercial), along with specialized variants for in/out-painting (Fill), edge-to-image (Canny), depth-to-image (Depth), image variation (Redux), and image editing (Kontext, Krea). The repository supports local installation with Python, optional TensorRT acceleration, and also provides an API for accessing Pro tier models and commercial licensing.

Key Features

Open-weight models with varying licenses (Apache 2.0 for schnell, non-commercial for dev variants)

Text-to-image generation with high-quality photorealistic output

Specialized models for inpainting, outpainting, structural conditioning (Canny edge, depth), image variation, and image editing

Minimal inference code for local deployment with Python

Optional TensorRT support for optimized performance

API access to Pro tier non-open-weight models via docs.bfl.ai

Commercial licensing available through bfl.ai/pricing/licensing with usage tracking

Pros & Cons

Pros

High-quality photorealistic output across multiple model variants
Open-weight models allow local deployment and fine-tuning
Includes models for diverse tasks beyond basic text-to-image (editing, in/out-painting, conditioning)
Apache 2.0 licensed schnell model for permissive use
TensorRT support enables optimized GPU inference
Commercial licensing path available for dev models

Cons

Dev models require a non-commercial license, limiting business use without additional licensing
Local setup requires significant GPU resources and Python environment configuration
No pre-built GUI or cloud deployment; inference code is minimal and requires coding knowledge
API access for Pro models is separate and may have usage costs

Best For

Creating photorealistic images from text promptsInpainting and outpainting to add or remove elements from imagesStructural conditioning: generating images guided by edge maps (Canny) or depth mapsImage variation: generating different versions of a given imageImage editing with contextual changes (Kontext) or via Krea modelResearch and experimentation with open-weight generative AI models

FAQ

What models are available in the Flux repository?

Flux offers FLUX.1 [schnell] (Apache 2.0), FLUX.1 [dev] (non-commercial), FLUX.1 Fill [dev] for in/out-painting, FLUX.1 Canny [dev] and Depth [dev] for structural conditioning (with LoRA versions), FLUX.1 Redux [dev] for image variation, FLUX.1 Kontext [dev] and Krea [dev] for image editing.

How can I install Flux locally?

Clone the repository, create a Python 3.10 virtual environment, and run 'pip install -e .[all]'. For TensorRT support, install from an NVIDIA PyTorch container and use 'pip install -e .[tensorrt]' with the extra index.

Can I use Flux models for commercial purposes?

FLUX.1 [schnell] is licensed under Apache 2.0 and can be used commercially. Dev models require a commercial license, which can be obtained through bfl.ai/pricing/licensing with a monthly usage fee.