MochiDiffusion

MochiDiffusion is a native macOS app that runs Stable Diffusion image generation entirely on your Mac. Instead of a Python stack running in a browser tab, it is a proper SwiftUI application that uses Apple’s Core ML framework and the Neural Engine, so generation is fast, efficient and completely local. The project has earned about 8,000 stars on GitHub and is actively maintained, with updates in 2026.

Overview

The app loads Core ML-converted Stable Diffusion models and generates images using the Apple Neural Engine (ANE), which keeps fans quiet and battery drain low compared with GPU-bound Python setups. It covers the workflows image creators expect: text-to-image generation with control over samplers, steps, guidance and seeds; image-to-image transformation; inpainting with a mask editor; and built-in upscaling for finished pieces. Because everything runs on-device, your prompts and images never leave the machine, and there is nothing to install beyond the app and the models you choose — no conda environments, no dependency conflicts. For Mac owners curious about generative art, it is one of the simplest on-ramps to Stable Diffusion that still exposes the controls serious users want.

Key Features

• Native macOS app — Stable Diffusion running on Core ML and the Neural Engine
• Text-to-image, image-to-image and inpainting workflows
• Built-in image upscaling for finished generations
• Fully local and private: no accounts, no cloud, no Python setup
• Fine-grained control of samplers, steps, CFG scale and seeds
• Works with the wide ecosystem of Core ML Stable Diffusion models

💳 Pricing

MochiDiffusion is free and open source under the GPL-3.0 license. You download your own models; generation on your Mac costs nothing.

Who Is It For

MochiDiffusion is for Mac users who want to generate images with Stable Diffusion in a polished native app, privately and offline, without wrestling with Python environments.