---
title: Lightweight LLM
emoji: 🪶
colorFrom: blue
colorTo: pink
sdk: docker
app_port: 7860
pinned: false
---

# Lightweight LLM Server

Self-hosted small LLM (Phi-3.5-mini, GGUF/Q4) served via llama.cpp for CPU inference.

- Endpoint: `POST /generate` with `{"prompt": "..."}`