A collection of 419 pre-trained AI models covering vision, speech, NLP and LLMs, all runnable from a shared Python CLI via the ailia SDK.
amd/RyzenAI-SW
Tutorials, demos and examples for running AI inference on AMD Ryzen AI PCs using the Ryzen AI Software tools and runtime.
- Stars
- 894
- Forks
- 135
- Open issues
- 105
- Last commit
- 2 months ago
What is it?
This repository is the examples, demos and tutorials companion to AMD's Ryzen AI Software, a toolkit for optimizing and deploying AI inference on AMD Ryzen AI powered PCs. It is a subset of the full Ryzen AI Software release, and installation instructions live in the official developer guide rather than in this repo.
The examples cover several model families: CNN image classification, object detection with YOLOv8, super-resolution, Transformer models such as a fine-tuned DistilBERT, and speech recognition with Whisper and NVIDIA Parakeet TDT. There are also LLM and vision-language model flows built on the ONNX Runtime GenAI API, including a RAG pipeline and GPT-OSS-20B inference, plus Windows ML samples for ResNet, Transformers and LLMs.
The repository uses Git LFS for large model files, so users need to install Git LFS and run git lfs pull after cloning. Each example directory has its own instructions. Quantization is covered by a tutorial using AMD Quark.
Key features
- 01Runs AI inference on AMD Ryzen AI NPUs and iGPUs
- 02Includes CNN, Transformer, object detection and super-resolution examples
- 03Provides LLM and vision-language model flows using ONNX Runtime GenAI
- 04Offers Windows ML samples for ResNet, Transformers and LLMs
- 05Includes speech recognition demos with Whisper and Parakeet TDT
- 06Shows model quantization with AMD Quark
Use cases
- Trying out local LLM or VLM inference on a Ryzen AI laptop
- Deploying an ONNX vision model on the Ryzen AI NPU
- Building a local RAG application with ONNX Runtime GenAI
- Adding on-device speech recognition with Whisper on Windows
Quick start
- 1
Install Git LFS from the official website
- 2
Set up Git LFS on your machine
git lfs install - 3
Clone the repository
git clone https://github.com/amd/RyzenAI-SW.git - 4
Pull the actual LFS files
git lfs pull
Things to know
- Targets only AMD Ryzen AI PCs, so the demos need Ryzen AI hardware.
- Some examples need large files via Git LFS, and installation follows external Ryzen AI Software docs.
- This repository is a subset of the Ryzen AI Software release, so each demo has its own setup instructions.
Topics
FAQ
- What is RyzenAI-SW?
- Tutorials, demos and examples for running AI inference on AMD Ryzen AI PCs using the Ryzen AI Software tools and runtime. This repository is the examples, demos and tutorials companion to AMD's Ryzen AI Software, a toolkit for optimizing and deploying AI inference on AMD Ryzen AI powered PCs. It is a subset of the full Ryzen AI Software release, and installation instructions live in the official developer guide rather than in this repo.
- Is RyzenAI-SW free to use? What license does it have?
- RyzenAI-SW is open source under the MIT license. Check the license terms before using it in commercial products.
- Is RyzenAI-SW still maintained?
- Updated within 6 months. The last commit to RyzenAI-SW was on Aug 18, 2026.
- What can I use RyzenAI-SW for?
- Trying out local LLM or VLM inference on a Ryzen AI laptop Deploying an ONNX vision model on the Ryzen AI NPU Building a local RAG application with ONNX Runtime GenAI Adding on-device speech recognition with Whisper on Windows
- How do I install RyzenAI-SW?
- Install Git LFS from the official website Set up Git LFS on your machine: git lfs install Clone the repository: git clone https://github.com/amd/RyzenAI-SW.git Pull the actual LFS files: git lfs pull
- Where can I get RyzenAI-SW?
- The source code and installation instructions are on GitHub: https://github.com/amd/RyzenAI-SW
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A toolkit for compressing, deploying, and serving LLMs and vision-language models, with TurboMind and PyTorch inference engines.
Rust-based LLM inference engine serving OpenAI- and Anthropic-compatible APIs, with quantization, multimodal models, and agent tooling.
A local voice assistant with a GLaDOS-style custom voice, camera vision, persistent memory, tools and a browser console.
A decision layer for LLMs and agents that routes requests by semantic similarity or a small decision model, returning calibrated confidence.
A llama.cpp fork focused on CPU inference speed, with extra quantization types and features for running GGUF LLMs locally.