OpenHuman is an open-source agent harness and desktop app with 42,000+ stars on GitHub, built to run personal AI agents — and whole fleets of them — with far less compute than typical agent tools.
Most agent harnesses run one heavy process per agent and resend a large prompt on every call, which makes always-on personal agents expensive. OpenHuman is written in Rust and engineered around efficiency: the project reports running more than 500 agents on a $10 VPS, cold starts of about a tenth of a second, and coding tasks finished in roughly 20 seconds in its own benchmark comparisons, using about 2.6 times fewer tokens than the typical agent tool. For people, it ships as a desktop app for macOS, Windows and Linux — a local-first personal AI with MCP support and a “second brain” orientation toward your own knowledge. For developers, the same core is a Rust library for embedding fleets of agents. Released recently and still in early beta, it became GitHub’s number-one trending repository for nine straight days after launch, and development is very active as of October 2026.
OpenHuman is free and open source under the GPL-3.0 license. You bring your own model API keys or local models, so costs are limited to the inference you actually use and the machine you run it on.
People who want a personal AI agent that lives on their own computer, and developers who need to run many agents at once without watching the token bill explode.