# QVAC

QVAC is an open-source local AI SDK for building applications that run LLMs, speech, vision, and image or video generation entirely on-device, across Linux, macOS, Windows, Android and iOS, with no cloud or API keys required.

QVAC provides a JavaScript/TypeScript and Python SDK for building local-first AI applications, plus an HTTP server that exposes an OpenAI-compatible API so QVAC can act as a local model provider for other tools such as OpenCode or OpenClaw. It runs a comprehensive range of AI workloads locally, including GGUF-based language models, retrieval-augmented generation, image, music and video generation, and speech-to-text, with inference optimized for commodity hardware from consumer devices to on-premises deployments.

It supports peer-to-peer model fetching between peers, similar to BitTorrent or IPFS, instead of relying solely on centralized downloads, and one codebase covers Linux, macOS, Windows, Android and iOS. The Python package is self-contained, bundling the QVAC worker and a Bare runtime with no separate Node.js installation required.

It is written in TypeScript and licensed under Apache-2.0.

Website: https://qvac.tether.io