SkillsApr 8, 2026·3 min read

Jan — Run AI Models Locally on Your Desktop

Open-source desktop app to run LLMs offline. Jan supports Llama, Mistral, and Gemma models with one-click download, OpenAI-compatible API, and full privacy.

SK
Skill Factory · Community
Quick Use

Use it first, then decide how deep to go

This block should tell both the user and the agent what to copy, install, and apply first.

  1. Download from jan.ai (Mac/Windows/Linux)
  2. Open Jan → Model Hub → Download a model (e.g., Llama 3.1 8B)
  3. Start chatting — fully offline, no API key needed
# Or use the local API (OpenAI-compatible)
curl http://localhost:1337/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model": "llama3.1-8b", "messages": [{"role": "user", "content": "Hello"}]}'

What is Jan?

Jan is an open-source desktop application for running AI models locally. It provides a ChatGPT-like interface that works completely offline. Download models with one click, chat privately, and expose a local OpenAI-compatible API for integration with other tools. Your data never leaves your machine.

Answer-Ready: Jan is an open-source desktop app for running LLMs locally. ChatGPT-like UI, one-click model downloads, OpenAI-compatible local API. Supports Llama, Mistral, Gemma, and GGUF models. Fully offline, complete privacy. 26k+ GitHub stars.

Best for: Developers and privacy-conscious users wanting local AI. Works with: Claude Code (as local backend), Cursor, any OpenAI-compatible tool. Setup time: Under 2 minutes.

Core Features

1. One-Click Model Download

Built-in model hub with curated models:

  • Llama 3.1 (8B, 70B)
  • Mistral 7B, Mixtral
  • Gemma 2
  • Phi-3
  • Any GGUF model from HuggingFace

2. OpenAI-Compatible API

from openai import OpenAI

client = OpenAI(base_url="http://localhost:1337/v1", api_key="not-needed")
response = client.chat.completions.create(
    model="llama3.1-8b",
    messages=[{"role": "user", "content": "Explain quantum computing"}],
)
print(response.choices[0].message.content)

3. Extensions System

  • TensorRT-LLM: NVIDIA GPU acceleration
  • Remote API: Connect to OpenAI/Anthropic as fallback
  • RAG: Local document Q&A

4. Cross-Platform

Platform GPU Support
macOS Apple Silicon (Metal)
Windows NVIDIA CUDA
Linux NVIDIA CUDA, Vulkan

Jan vs Alternatives

Feature Jan Ollama LM Studio
GUI Full desktop app CLI only Full desktop app
API OpenAI-compatible OpenAI-compatible OpenAI-compatible
Extensions Plugin system Limited No
Open source Yes (AGPL-3.0) Yes No
Model format GGUF GGUF, safetensors GGUF

FAQ

Q: What hardware do I need? A: 8GB RAM minimum for 7B models. 16GB+ recommended. Apple Silicon Macs work great with Metal acceleration.

Q: Can I use it as a backend for Claude Code or Cursor? A: Yes, Jan exposes an OpenAI-compatible API at localhost:1337. Point any tool that supports custom endpoints to it.

Q: Is it truly private? A: Yes, everything runs locally. No telemetry, no data collection. You can verify — it is open source.

🙏

Source & Thanks

Created by janhq. Licensed under AGPL-3.0.

janhq/jan — 26k+ stars

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.