How-to

How to run a local coding assistant on a Mac

Last updated 2026-11-04

What you need first for “how to run a local coding assistant on a mac”

Apple Silicon Mac, macOS 15 or later, and for how to run a local coding assistant on a mac the unified-memory minimum that the chosen tier requires (6 GB for Nano, 12 GB for Lite, 16 GB for Quick (24 GB to run it fully in memory), 24 GB for Core and Vision 3.8, 64 GB for Plus). For run a local coding assistant on mac, those are the minimums to load a tier; a long agent session needs more. You need the internet to download each model (and once to activate a Pro key); after that, chat works offline.

Steps

1. Install Outlier. Per the install guide. macOS 15+ Apple Silicon, 24 GB RAM minimum for Core.
2. Switch to Code mode. The mode toggle is in the chat composer. The system prompt is tuned for terse code-first responses.
3. Open a project. Use the project chip to scope context to a folder.
4. Ask for a focused review. Quote a single function. Bigger asks blow the context budget on smaller tiers.

What is the specific thing to know about this guide?

Code mode was renamed from Agent mode. The system prompt is tuned for code-first responses, default temperature is lowered, and the autonomy mode toggles default to manual approval.

Core defaults to a 32K-token context window and goes up to 128K on Pro, so a whole file plus its imports usually fits without truncation. It needs 24 GB of unified memory. Below that, drop to Quick and scope each question to a single function rather than a whole file.

What can go wrong with this guide?

• The coding agent is a Pro feature: the free tiers (Nano and Lite) are chat only. For the agent, and for top code quality, you need Pro, which includes Core.
• Project chip context can blow the window if the folder is large.

How does this guide differ from the cloud equivalent?

Hosted coding assistants stream from a remote endpoint that your code went to. Outlier streams from a local sidecar at 127.0.0.1:8766.

What does this guide not claim about “how to run a local coding assistant on a mac”?

This guide does not claim feature parity with cloud-side workflows for “how to run a local coding assistant on a mac”. Specifically, the product surface covers chat, file attachment, the local agent loop, project scoping, and the model picker. Cross-device sync, team workspaces, and shared session history are out of scope and are not on the v1.9 backlog either.

What is different about Code mode versus chat?

Code mode ships with a terser system prompt, a lower default sampling temperature, and a default project-scope chip. The agent loop has manual approval on as the default; auto-approve and sandboxed modes are settings, not the default. The reason that distinction is non-trivial is that the agent loop can run shell commands; manual approval is the safe choice when starting out.

Where does this guide fit in the rest of the lineup?

Code mode pairs naturally with the project chip. A typical session opens a single file, asks for a refactor or test scaffold, then iterates inline; the chat history persists across mode switches.

One unique number

4 steps. Once the model is downloaded your chats and files stay on your Mac. Core is one model; the separate Code entry that shared its safetensors has been folded back in.

Download Outlier for Mac

On your phone? Send the Mac link to yourself

Free · M1 or newer · macOS 15+ (free update) · 276 MB. Intel Macs are not supported.