# Local LLM Deployment Guide — llama.cpp Sovereign Inference

llama.cpp provides sovereign, offline-capable inference on CPU, CUDA, and
Apple Silicon. One binary, one model path, one health endpoint.

## Quick Start

    git clone https://github.com/ggerganov/llama.cpp.git
    cd llama.cpp && cmake -B build && cmake --build build --config Release -j$(nproc)
    sudo cp build/bin/llama-server /usr/local/bin/
    mkdir -p /opt/models/llama
    wget -O /opt/models/llama/Qwen2.5-7B-Instruct-Q4_K_M.gguf "https://huggingface.co/Qwen/Qwen2.5-7B-Instruct-GGUF/resolve/main/qwen2.5-7b-instruct-q4_k_m.gguf"
    llama-server -m /opt/models/llama/Qwen2.5-7B-Instruct-Q4_K_M.gguf --host 0.0.0.0 --port 11435 -c 4096 -t $(nproc) --cont-batching
    curl http://localhost:11435/health

## Model Path Convention

- /opt/models/llama/ — Production (system-wide)
- ~/models/llama/ — Per-user (dev)
- MODEL_DIR env var — Override

## Recommended Models

- Qwen2.5-7B-Instruct (4.7GB, 8GB RAM, 25-40 tok/s) — Fleet standard
- Qwen2.5-3B-Instruct (2.0GB, 4GB RAM, 50-80 tok/s) — VPS Beta
- Mistral-7B-Instruct-v0.3 (4.4GB, 8GB RAM) — Alternative

## Quantization Guide

- Q6_K (5.5GB) — Best quality/speed, RAM > 12GB
- Q4_K_M (4.7GB) — Fleet standard
- Q3_K_M (3.4GB) — < 6GB RAM fallback

## Hardware Targets

- VPS Beta (2 vCPU, 4GB): Qwen2.5-3B-Q4_K_M, ctx 2048, ~40-60 tok/s
- VPS Alpha (4 vCPU, 8GB): Qwen2.5-7B-Q4_K_M, ctx 4096, ~20-35 tok/s
- Mac Apple Silicon: Qwen2.5-7B-Q6_K, Metal, ~30-50 tok/s

## Health Check

    curl -sf http://localhost:11435/health
    curl -s http://localhost:11435/v1/models

## API Compatibility

llama-server exposes OpenAI-compatible API at /v1/chat/completions.

## Troubleshooting

- Won't start: use smaller model or lower quant
- Slow: match -t to available cores
- OOM: reduce -c context size
- Port in use: lsof -i :11435