mflux-streamlit

June 5, 2026 · View on GitHub

PyPI - Version PyPI - Python Version

A web app gui for mflux python library implemented with Streamlit

Table of Contents

Features

  • Multiple MFLUX model families from a single UI, each with its own variants and controls:
    • FLUX.1 (Schnell / Dev / Krea-Dev) — txt2img, img2img, Kontext editing, ControlNet (Canny)
    • FLUX.2 (Klein 4B/9B, distilled & base) — fast txt2img and image editing
    • Z-Image (Turbo / base) — fast, small, high realism
    • FIBO (FIBO / FIBO-lite / FIBO-Edit) — natural-language or JSON prompts, editing
    • Qwen-Image (txt2img / edit) — strong prompt understanding and world knowledge
    • SeedVR2 (3B / 7B) — image upscaling by target resolution or scale factor
    • Depth Pro — fast, accurate depth-map estimation
  • Multiple LoRAs and scales for each LoRA adapter (where supported)
  • Optional reference / source image for img2img, editing, ControlNet, upscaling and depth
  • Per-variant quantization (4-bit / 8-bit), guidance, steps and negative-prompt controls

Previews

Left: FLUX.1Right: FLUX.1 with LoRA
FLUX.1 SchnellFLUX.1 Schnell with LoRA
Left: FLUX.1Right: FLUX.1 with LoRA
ControlNet Image
Source		FLUX.1 Schnell with Controlnet

Todo List

  • Beautify the web UI and improve UX

Installation

Normal User Install

brew install uv
uv tool install mflux-streamlit
cd /your/directory

mflux-streamlit

Developer Install

# dev workaround: run the main.py
uv venv && source .venv/bin/activate
uv pip install -e .
streamlit run src/app.py

License

mflux-streamlit is distributed under the terms of the MIT license.