DEDUPED Hot AI repos. Real use cases. What people build. Get the Monday 5

How to chat with an AI model fully offline with Jan

Install Jan, download an open model from Hugging Face and chat without sending anything to the cloud.

Short answer: Download Jan for your OS from Jan.ai or GitHub Releases: Jan.exe, Jan.dmg, Jan.deb or Jan.AppImage; then open Jan and download a model such as Llama, Gemma or Qwen from Hugging Face; pick a model size your RAM can handle, for example 3B on an 8GB Mac. All steps.

Who it is forSomeone who wants ChatGPT-style help with private notes or documents without a subscription.

The problemCloud chatbots send every message to a third party, and setting up local models usually takes a terminal.

The outcomeA desktop chat app running an open model locally, with everything kept on your machine.

What you need first

  • Windows 10+, macOS 13.6+ or most Linux distributions
  • RAM for the model size: 8GB for 3B models, 16GB for 7B, 32GB for 13B on macOS
  • To build from source: Node.js 20+, Yarn 4.5.3+, Make and Rust

Steps

  1. Download Jan for your OS from Jan.ai or GitHub Releases: Jan.exe, Jan.dmg, Jan.deb or Jan.AppImage
  2. Open Jan and download a model such as Llama, Gemma or Qwen from Hugging Face
  3. Pick a model size your RAM can handle, for example 3B on an 8GB Mac
  4. Start chatting; everything runs locally when you choose a local model

Prompt for your agent

Help me set up janhq/jan on my <OS> computer with <RAM> of memory. Recommend a local model size that will run smoothly and walk me through downloading it and starting a chat.

Caveat. Speed and quality depend on your hardware; larger models need much more RAM.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for janhq/jan and "Chat with an AI model fully offline"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer