Block just open-sourced mesh-llm, a peer-to-peer system that lets anyone pool spare GPU compute to run large open-source AI models without relying on any cloud provider. If a model fits on your machi...