Claude Desktop · rapid-mlx 0.16.6

Claude Desktop with a Local Model on Mac

Connect Claude Desktop to a local model on your Mac through its in-app Gateway settings.

Set up the local Gateway

Install Rapid-MLX, then run:

$ rapid-mlx launch claude-desktop --model qwen3.5-4b-4bit --port 7659

This command saves Rapid's private model-tier mapping and prints a Gateway URL and API key. It does not start the server or edit Claude Desktop. Start the matching model on that port with the saved key, and keep it running while using Claude Desktop:

$ RAPID_MLX_API_KEY="$(cat ~/.rapid-mlx/connect/api-key)" rapid-mlx serve qwen3.5-4b-4bit --port 7659
  1. In Claude Desktop, open Help → Troubleshooting → Enable Developer Mode.
  2. Open Developer → Configure Third-Party Inference → Gateway.
  3. Paste the URL and API key printed by Rapid, choose the default Bearer authentication scheme, and select Test.
  4. Select Apply Changes.

Next day: Start the same server again, then use Claude Desktop. Re-run rapid-mlx launch claude-desktop --model qwen3.5-4b-4bit --port 7659 if you need the Gateway details for this example. Use the same model and port as your server. Claude Desktop's settings change only when you apply them in its own window. rapid-mlx launch claude-desktop --remove disables Rapid's tier mapping.

Model by RAM: For a starting model, use qwen3.5-4b-4bit on a 16 GB Mac, qwen3.5-9b-4bit on an 18–23 GB Mac, and qwen3.8-27b-4bit on 32 GB or more. On an 8–15 GB Mac, the installer suggests lfm2.5-2.6b-4bit; small models may struggle with multi-step coding tasks. See the RAM guide. Use a tool-capable model that fits your Mac. First reply: The first reply can take longer than later turns: an uncached model must download, the server must load it, and the agent may send a long prompt that takes time to prefill. Wait for the model to become ready; there is no fixed first-reply time. A warm, cached model is usually quicker.

Quick answer

How do I use Claude Desktop with a local model on Mac? Run rapid-mlx launch claude-desktop --model qwen3.5-4b-4bit --port 7659, start the matching Rapid-MLX server, then paste the printed Gateway URL and key into Claude Desktop's in-app Gateway settings.

See coding agents for the one-command terminal launch flow.