SwarmLLM

SwarmLLM is a ChatGPT alternative for people who want to run AI models on their own hardware. It can use a single computer or combine several machines to handle a model that won't fit on one.

• Local chat: A browser interface shows which models fit your computer and lets you download models from Meta, Google, Alibaba and Mistral.
• Shared inference: Link your laptop, desktop and home server, or join a public swarm over the internet, with each computer holding part of a model.
• App connections: OpenAI-compatible and Anthropic Messages APIs let chat apps, AI agents and coding assistants, including Claude Code, use it as a backend; it also provides an MCP server.
• Optional cloud routing: You can connect external model providers with your own keys, which sends those requests to the chosen services.

Conversations stay on your machine when the model runs entirely locally. Public swarm connections are encrypted, but participating computers can see what they process, so public inference isn't suitable for sensitive conversations. The swarm handles medium-sized models, and nearby computers respond faster than distant ones. Your computer can also contribute spare capacity to other users.

The app is free and open source under Apache 2.0, with no account required. It runs on Windows, Linux and Apple Silicon Macs; 8 GB or more memory is recommended, and a GPU isn't required. Windows supports NVIDIA, AMD and Intel GPUs for local inference. Distributed GPU inference on Windows and GPU inference on Linux require an NVIDIA RTX 30-series or newer. Apple Silicon Macs use the CPU.