Inference Snaps¶
Inference snaps are generative AI models that are packaged and optimized for efficient performance on local hardware. They automatically detect your host machine’s hardware and install runtime and model weight optimizations that best match its capabilities.
Inference snaps provide an automatic, unified detection and installation mechanism. This results in model and runtime choices that make effective use of your CPU, GPU, or NPU.
With inference snaps, you get fast, local AI capabilities for applications such as IDEs, chat bots, and media editing tools. They also provide application developers with a standardized and reliable local API to integrate AI features into their software without the need for complex hardware-specific tuning.
In this documentation¶
Get started - a hands-on introduction for new users.
Step-by-step guides covering key operations and common tasks
Technical information - specifications, user manuals, and architecture
Concepts - discussion and clarification of key topics
Project and community¶
The Inference Snaps project is a member of the Ubuntu family. It is open source and warmly welcomes community projects, contributions, suggestions, fixes and constructive feedback.