The most useful open models today, like Qwen 3.8 at 27 billion parameters, can't run on your phone. You need a big PC, a workstation, or someone's cloud. Meanwhile most of us have a phone, a laptop, maybe an old PC, all sitting around with memory and GPUs doing nothing. SwarmLLM pools them together to put them to use: every device opens a link, takes a slice of the model in its browser tab, and as a group they run the whole thing. Peer-to-peer inference with anyone, anywhere.