Get started with Stud

Your workshop is ready.

Written for 1.0.0-preview.8. Available download: 1.0.0-preview.8 · Public preview.

Stud works with your existing AI app. Describe your project in ChatGPT for desktop, then use Stud to inspect the 3D model, refine the details, and plan your materials. The setup below uses the Codex workspace inside ChatGPT for desktop.

1. Download for your computer

Choose your installer. On a Mac, open Apple menu → About This Mac: an Apple M-series chip needs Apple Silicon; an Intel processor needs Intel. Windows installers are for x64 computers. There is no Linux desktop installer.

Stud’s installer includes its modeling runtime, Git, and viewer. You don’t need to install Python or Node.js separately. You will need ChatGPT for desktop installed for the guided workflow below; Stud does not include an AI subscription.

2. Connect Stud

macOS

  1. Open the DMG and drag Stud into Applications.
  2. Open Stud and choose Connect CLI and skill. macOS may request an administrator password to connect the command-line tool.
  3. In ChatGPT for desktop, open Codex and start a new task so it loads the bundled Stud skill. Ask it to run stud --version to confirm the connection.

If you see Repair connections, use it to reconnect the tools to this installation. If Stud reports a conflict with a custom installation, back up and move that custom installation aside first.

Windows

  1. Run the -setup.exe installer. It installs for your user account.
  2. Copy the entire bundled skills/stud-design folder from Stud’s installation directory into your skills directory (usually ~/.codex/skills, or the skills directory under your custom CODEX_HOME).
  3. Restart ChatGPT for desktop and any open terminals, then start a Codex task and ask it to run stud --version.

If the command isn’t found, open Stud and choose Install command-line tool to repair the PATH entry, then restart ChatGPT for desktop.

3. Start your first project

In ChatGPT for desktop, give your project a real size and a purpose. Try the workbench tutorial, or start with this:

Use Stud to design a workbench for my garage, 6 feet wide and 24 inches deep, with a plywood top and a shelf underneath. Create a new project folder, open the viewer, and show me the dimensions and material list.

Keep your project in its own folder, separate from the app. The agent starts the local viewer and can open it in a browser. Stud’s modeling and exports run locally; your AI assistant has its own connection requirements.

When something doesn’t work

ChatGPT can’t find Stud

Check the connection in the desktop app, then restart ChatGPT for desktop. Ask it to run stud --version. Use the CLI path shown by the app if a restarted process still does not see the command.

The viewer shows my old design

A failed rebuild keeps the last good preview. Ask the agent to inspect the build error and correct it. Review the new design only after a successful rebuild.

The viewer address is already in use

Ask the agent to stop the other viewer or serve this project on another port, such as 8766, and open the new local address.

I want to reopen or back up a project

Use Add project in the desktop app to select an existing project folder. Only open projects you trust: their design files execute Python. Back up the entire folder, including hidden files, to retain history and artifacts.

Keep making.

Explore the tutorialsDesktop reference for this release