stable-diffusion.cpp and ADI-Stable-Diffusion
These are competitors offering alternative C/C++ implementations for running diffusion model inference via ONNX Runtime, with the first supporting a broader range of model architectures (SD, Flux, Qwen) while the second emphasizes cross-platform acceleration as a unified framework.
About stable-diffusion.cpp
leejet/stable-diffusion.cpp
Diffusion model(SD,Flux,Wan,Qwen Image,Z-Image,...) inference in pure C/C++
Built on ggml with zero external dependencies, it supports diverse architectures—image generation (SD1.x through FLUX.2), video (Wan2.1/2.2), and editing models—alongside LoRA, ControlNet, and quantized GGUF weights. Multi-backend acceleration includes CUDA, Vulkan, Metal, and OpenCL, with optimizations like Flash Attention, VAE tiling, and latent consistency models for efficient CPU/edge deployment.
About ADI-Stable-Diffusion
Windsander/ADI-Stable-Diffusion
Accelerate your Stable Diffusion inference with the library's universal C/C++ framework design, powered by ONNXRuntime & across platforms.
Provides both a C++ library and CLI tool that decouples inference from the Stable Diffusion framework by converting models to ONNX format, enabling flexible hardware acceleration (CUDA, TensorRT, CoreML, NNAPI) through configurable ONNXRuntime providers. Supports multiple inference modes including text-to-image and image-to-image with fine-grained control over scheduler algorithms, noise prediction strategies, and tokenization methods. Built with cross-platform deployment in mind via automated build scripts for macOS, Windows, Linux, and Android with configurable compile options for package size optimization.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work