drmihirbrahme/midge: Tiny engine, immense models — run large MoE LLMs (gpt-oss, Mixtral, Qwen3-MoE) on ordinary machines by streaming experts from disk. OpenAI-compatible server with tool calling + hybrid cloud relay; CPU, Apple Silicon & CUDA (MLX).

Source-linked topic cluster with 1 signals across related articles, projects, models, papers, and source updates.