hugpy
inference you own

Self-hosted, GPU-pooling LLM inference — pool a workstation GPU, an idle laptop, even a phone, behind one streaming, OpenAI-compatible API.

A model registry with on-demand load/unload, streaming chat, Whisper transcription, embeddings & summarization — 40+ models from a single console, on hardware you own.

GitHub· PyPI· npm· X· Instagram· Threads· Reddit· Minds © 2026 · source-available