Skip to content

mflux

Model Serving & MLOps

A minimal, line‑by‑line MLX port that runs state‑of‑the‑art generative image models locally on macOS.

Python Latest v.0.18.0 · 1mo ago Security brief →

Features

  • Quantization and local model loading
  • LoRA support (multi‑LoRA, scales, library lookup)
  • Metadata export & reuse with prompt file handling
  • Text‑to‑image and image‑to‑image generation
  • ControlNet conditioning, depth estimation, upscaling

Recent releases

View all 43 releases →
No immediate action
v.0.18.0 Mixed

Model support + KV‑cache + bug fixes

v.0.17.5 Bug fix

Fixes Qwen Image Edit saving where VisionTransformer weights were omitted.

v.0.17.4 Bug fix

Fixed Z-Image PEFT/ModelScope LoRA key mapping with `.default` tensor variants for correct adapter resolution.

v.0.17.3 Bug fix

FLUX.2 edit guidance metadata now correctly preserves and reports actual guidance values instead of always showing 1.0.

v.0.17.2 Bug fix

Fixed tokenizer cache resolution for partially cached Hugging Face repositories.

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

About

Stars
2,206
Forks
160
Languages
Python Makefile

Install & Platforms

Platforms
macos

Beta — feedback welcome: [email protected]