Run Continue with a Local Model on Your Mac

Continue is the open-source AI assistant for VS Code, JetBrains and the terminal. One model entry in its config points it at a model served on your own Mac.

Continue puts chat, inline edits and an agent into VS Code, JetBrains and a terminal CLI (cn). Every model it uses is described in one config file, and a model can be any OpenAI-compatible server, including Rapid-MLX on your Mac.

With a local model, the code Continue reads and the prompts you write stay on your machine, and there's nothing to pay per token.

The whole thing

1. Install Rapid-MLX and serve a model.
2. Add a Rapid-MLX model to ~/.continue/config.yaml.
3. Pick it in Continue.

1 · Install Rapid-MLX

curl -fsSL https://rapidmlx.com/install.sh | bash

Prefer Homebrew? brew install rapid-mlx. Continue itself comes from your editor's extension marketplace, or npm install -g @continuedev/cli for the terminal.

2 · Serve a model

rapid-mlx serve qwen3.5-4b-4bit

That's the engine's pick for a 16 GB Mac and the model we verified with. rapid-mlx recipe prints the pick for your Mac's memory, and Best local LLMs by Mac has measured speeds per chip. Keep this terminal open.

3 · Add the model to Continue

Continue reads ~/.continue/config.yaml. Add a model with the openai provider and your local server as its apiBase:

name: Rapid-MLX local
version: 1.0.0
schema: v1
models:
  - name: Rapid-MLX
    provider: openai
    model: qwen3.5-4b-4bit
    apiBase: http://localhost:8000/v1
    apiKey: not-needed
    roles:
      - chat
      - edit
      - apply

The model value is the alias you served. If the file already lists other models, add this entry to the same models: list.

4 · Use it

In VS Code or JetBrains, pick Rapid-MLX in Continue's model dropdown and chat. In a terminal, cn picks up the same file:

# inside your project folder
cn -p "Add a function add(a, b) to hello.py that returns a + b. Edit the file."

From our run:

I've added the `add(a, b)` function to `hello.py`. The file now contains:

- `greet(name)` - returns "Hello, " + name
- `add(a, b)` - returns a + b

Frequently asked questions

Can Continue use a local LLM on a Mac?

Yes. Add a model with provider: openai and apiBase: http://localhost:8000/v1 to ~/.continue/config.yaml, serve it with Rapid-MLX, and pick it in Continue. Chat, edit and the agent then run on your Mac.

Is Continue with a local model private?

Requests go from Continue to localhost, so your prompts and code stay on the Mac. Continue's own optional features, such as signing in to Continue Hub, follow Continue's policies; the local model path doesn't need them. Rapid-MLX itself sends anonymous usage counts by default, never prompts or code; rapid-mlx telemetry off turns that off (details).

Which model should I use with Continue?

rapid-mlx recipe prints the engine's pick for your memory. For the agent and multi-file edits, use the largest that fits, such as qwen3.8-27b-4bit on 32 GB or more.

Where to go next

• Continue reference: the launch and agents commands in detail.
• Best local LLMs by Mac: measured speed for your chip and memory.
• Rapid-MLX vs mlx-lm: the same weights, measured task by task.
• All agent guides: every agent Rapid-MLX configures.

curl -fsSL https://rapidmlx.com/install.sh | bash