llm-nest-python 0.1.0

本地 LLM Runtime 与 GGUF 模型管理工具

llm-nest

English | 中文

Local LLM runtime and GGUF model management tool. Think of it as ollama for your terminal — lightweight, offline-first, no daemon required.

Features

• Model management — search, download, list, and delete GGUF models from HuggingFace Hub
• Interactive chat — REPL mode with multi-turn conversation and streaming output
• API server — OpenAI-compatible /v1/chat/completions endpoint with SSE streaming

Install

pip install llm-nest-python

With API server support (FastAPI + uvicorn):

pip install llm-nest-python[server]

From source (uv)

git clone https://github.com/NingCold/llm-nest-python.git
cd llm-nest-python
uv sync
uv run llmn --help

Quick Start

Download a model

llmn hub get tensorblock/tinyllama-15M-GGUF -f tinyllama-15M-Q4_K_M.gguf

List local models

llmn model list

Run interactive chat

llmn run tinyllama-15M-Q4_K_M

Loading model: tinyllama-15M-Q4_K_M (Q4_K_M, 0.0GB) ...
Type 'exit' or press Ctrl+C to quit

You: Hello
Assistant: Hello! How can I help you?

You: exit
Bye!

Single-shot inference

llmn run tinyllama-15M-Q4_K_M -p "What is 2+2?" --max-tokens 50

With system prompt

llmn run tinyllama-15M-Q4_K_M -s "You are a helpful assistant."

API Server

Start an OpenAI-compatible HTTP server:

llmn serve tinyllama-15M-Q4_K_M --port 8000

Endpoints

Endpoint	Method	Description
/health	GET	Health check
/v1/models	GET	List loaded models
/v1/chat/completions	POST	Chat completion (stream + non-stream)

Usage with curl

# Non-streaming
curl http://127.0.0.1:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"messages": [{"role": "user", "content": "Hello"}]}'

# Streaming (SSE)
curl http://127.0.0.1:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"messages": [{"role": "user", "content": "Hello"}], "stream": true}'

Usage with OpenAI Python SDK

from openai import OpenAI

client = OpenAI(base_url="http://127.0.0.1:8000/v1", api_key="none")
resp = client.chat.completions.create(
    model="tinyllama-15M-Q4_K_M",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

Commands

llmn model list                          # List local models
llmn model search <query>                # Search local + Hub models
llmn model info <name>                   # Show model details
llmn model remove <name>                 # Delete a local model

llmn hub search <query>                  # Search HuggingFace Hub
llmn hub get <repo_id> [-f filename]     # Download model from Hub

llmn run <model> [-p prompt] [-s system] # Interactive chat or single-shot
llmn serve <model> [--port 8000]         # Start API server

llmn lang zh                             # Switch to Chinese
llmn lang en                             # Switch to English
llmn version                             # Show version

Storage

Models are stored in ~/.llmn/models/. Config is at ~/.config/llm-nest/config.json.

Development

uv sync
uv run pytest
uv run ruff check .
uv run pyright

License

MIT