mesh-llm turns spare compute into a peer-to-peer inference cloud for open models.
Sizes DeepSeek-V4-Flash's maker never released: distills, merges and other community models named after it.
