A developer has released ClusterChat, an open-source distributed chatbot enabling a 27B parameter model (Qwen3.8-27B q8) to run across consumer GPUs entirely in the browser. The system partitions the LLM into block ranges, allowing multiple users to collaboratively host model shards that exceed individual hardware capacity, demonstrated on an RTX 3090 and GTX 1070 Ti. Unlike similar projects such as exo or Petals, ClusterChat operates natively within the browser environment.