Skip to content

The ultimate harness for coding agents and beyond. All your agents. All your machines. One command center. Start with code, then follow your curiosity and build across disciplines: CAD, circuits, robots, games and music.

1.6 KB
Raw Blame

Download and start a local model in hn

A local model runs on your own computer instead of a provider's servers.

1. Install hn

In a terminal, run:

curl -fsSL https://harness.autonomous.ai/cli/install.sh | bash

Then type hn and press Enter. The first time, hn signs you in.

2. Open a harness

Open the harness that will use the local model: press Ctrl+B, then Shift+N, choose an agent with the Up and Down arrow keys and press Enter. Claude Code, Codex, OpenCode can use local models.

Choosing an agent for a new harness

3. Open Models

In the harness, press Ctrl+B, then Shift+I. The Models screen opens for that harness.

If you see Set up local & shared models, press Enter on it first.

Models, opened for the harness

4. Choose a model

Get for this computer is the model catalog: models that fit this computer's memory (on NVIDIA, its GPU's VRAM), ranked for coding. Each row shows the download size and estimated speed.

Use the Up and Down arrow keys to select a model, then press Enter twice.

A model selected, with its details on the right

5. Wait

hn downloads the model, starts it and moves your harness onto it. Downloads are often 10–20 GB. If one stops, press Enter twice on the model again to resume it.

Downloading

6. Done

Your agent now answers with the local model.

The agent answering with the local model