LocalAGI is an open-source, self-hostable AI agent platform with 2,000+ stars on GitHub that runs advanced agents entirely on your own hardware — a complete drop-in replacement for OpenAI’s Responses API with no cloud in the loop.
LocalAGI starts from a privacy premise: agent capabilities should not require sending your data to someone else’s platform. It provides a full agent platform — agents that can plan, use tools and hold state — served through an API compatible with OpenAI’s Responses API, so applications written against that interface can be pointed at your own deployment instead. It is designed for consumer-grade hardware, running on CPU or GPU, and pairs naturally with local model serving from the same ecosystem of local-AI projects. The result is an agent stack you can run in a homelab, on a workstation, or inside an air-gapped environment where external AI services are not an option. The project is MIT licensed, from the author of LocalAI, and actively maintained with updates through October 2026 — an important signal in a category full of abandoned experiments.
LocalAGI is free and open source under the MIT license. Your costs are the hardware and electricity it runs on; there is no subscription and no per-request platform fee.
Privacy-focused developers, homelab builders and organisations in restricted environments who want real agent capabilities — API-compatible with the OpenAI-style ecosystem — without any data leaving their own machines.