Install Jan & download your first local model
Get a private ChatGPT-style assistant running in under 20 minutes
Jan is a free, open-source (Apache-2.0), offline-first desktop app for Mac, Windows, and Linux — a ChatGPT-style assistant whose models run 100% on your own machine, so no data ever leaves it. There is no account and no command line: you install the app, pick a model from a built-in Hub, and chat. Models are open-weight LLMs (families like Llama, Gemma, Qwen, and GPT-oss) — a smaller model downloads faster and runs on a modest laptop, while a larger model gives better answers but needs a more capable machine. When in doubt, start with the smallest model the Hub recommends for your hardware.
- 1 Download Jan at jan.ai — choose the installer for your platform (Mac, Windows, or Linux). Run it; no account is required.
- 2 Open the model Hub inside the app. Browse the curated list and pick a small open-weight model to start (a Llama, Gemma, or Qwen variant is a good first choice). Click to download it — the file downloads straight to your machine.
- 3 Wait for the download to finish. Open-weight models are multi-gigabyte files, so this can take a few minutes depending on the model and your connection.
- 4 Start a new chat, select your downloaded model, and type:
Explain RNA-seq in two sentences for a biologist who has never heard of it. - 5 Read the reply. If it arrives — even slowly — your setup is working. The model ran entirely on your own machine.
You received a coherent reply from a local model inside Jan, generated entirely on your own machine with nothing sent to the cloud.
Find the smallest model that answers well enough for your work
Bigger is not always better when your laptop's memory is the constraint. The goal is the smallest model that gives you answers you can trust for your actual tasks.
Download a second model from the Hub at a different size or from a different family, ask both the same science question, and decide which is your daily driver.
- 1 In the Hub, pick a second model — either a smaller variant of the same family or a different family entirely (Llama, Gemma, Qwen, GPT-oss).
- 2 Download it and chat with it using the same prompt you used in the USE phase.
- 3 Compare: response quality and response speed on your hardware.
- 4 Pick a winner and note why — quality or speed.
A one-sentence verdict: which model you chose and the reason (quality or speed).