The latest Data Phoenix news tagged GPU pooling.
iroh launched Mesh LLM, an 18 MB tool that pools GPUs across machines into a single OpenAI-compatible API endpoint for models up to 235B parameters.