Petals Launches Peer‑to‑Peer Platform for Running Large Language Models at Home

Petals introduced a distributed system that mimics BitTorrent architecture. The network enables individuals to execute large language models locally. Users contribute compute resources to a shared pool. Model inference is performed across participating devices in real time. The approach reduces reliance on centralized cloud services. Developers can access the platform through the petals.dev website. Early adopters report lower latency for certain model tasks. The project aims to democratize access to powerful AI capabilities.