CNChubromania/flux2-klein-base-9b-fp8-demo

⚡ FLUX.2 [klein] Base 9B FP8 Studio

Interactive demonstration and repository for Black Forest Labs' FLUX.2 [klein] Base 9B FP8.

Overview

• —Model: black-forest-labs/FLUX.2-klein-base-9b-fp8
• —Architecture: Next-generation Rectified Flow Transformer quantized in FP8 (8-bit floating point).
• —Conditioning: Qwen 3 8B text encoder with support for up to 12,288 tokens.
• —Serving code: Included in app.py for ZeroGPU and Gradio.
• —Dependencies: Listed in requirements.txt.

Running Locally / With Hardware

You can run the full PyTorch + Gradio app directly on a GPU with:

pip install -r requirements.txt
python app.py

[!NOTE] black-forest-labs/FLUX.2-klein-base-9b-fp8 is a gated model. To download the model weights, make sure to accept the license agreement on the model card and provide your HF_TOKEN.