Run Mistral Locally on Your MacBook Pro

Mistral is the European open-weights family known for punching above its size — Mistral Small 24B is a strong all-rounder that fits a 24 GB Mac, Devstral is tuned for coding, and the 119B flagship runs on a Studio. Here's the size that fits your Mac, and the one command that serves it.

Mistral is the open-weights family with a reputation for efficiency — models that punch above their parameter count, released under permissive licenses. On a Mac, the sweet spot is Mistral Small 24B: a strong general-purpose model that fits a 24 GB machine. There's a coding-tuned Devstral in the same size class, and a 119B flagship for Studios. All of it runs natively on Apple Silicon.

This guide is the short path: pick the right Mistral for your Mac's RAM, serve it with one command, and point your tools at it.

The 60-second version

curl -fsSL https://rapidmlx.com/install.sh | bash   # or: brew install rapid-mlx
rapid-mlx serve mistral-24b-4bit                    # pick your size below

Then talk to it — it's a standard OpenAI-compatible server on http://localhost:8000:

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model": "default", "messages": [{"role": "user", "content": "Write a haiku about garbage collection, then explain the reference."}]}'

First run downloads the weights; after that, startup is seconds. No account, no API key, nothing leaves your machine.

Which size fits your Mac

Sizes below are what the weights occupy on disk (and roughly in unified memory once loaded). Leave headroom for your context window and the rest of macOS.

Not sure where you land? The hardware tier guide maps every Mac from an 8 GB Air to a 512 GB Studio to a recommendation, and models.rapidmlx.com lets you filter the whole catalog by your RAM.

Tool calling works out of the box

serve auto-configures the right chat template and tool-call parser for every Mistral alias, and grammar-constrained tool calling is on by default — so function calls come back as valid JSON, not as a paragraph that almost parses. That's the boring-but-load-bearing part: Mistral-on-your-Mac slots into tools that expect OpenAI semantics without any glue code.

Wire it into your editor or agent

The server speaks the OpenAI protocol, so anything with a "custom base URL" box can use it:

• Claude Code — run the CLI

agent fully local against your Mistral.

• Cursor — chat, Ctrl-K and

composer against http://localhost:8000/v1.

• Codex CLI and

Aider — both speak the same OpenAI-compatible endpoint.

For coding and agent work, Devstral is the one to reach for; for general assistance, Mistral Small 24B is a strong, efficient default. Both serve through the same endpoint, so switching is one command.

How fast is it?

We publish real numbers rather than vibes: our five-model MacBook Pro benchmark puts local models through coding, reasoning, tool-calling and memory tests, and the M3 Ultra benchmark covers the desktop end. Mistral Small 24B is efficient for its capability — comfortably interactive on a 24 GB MacBook.

Frequently asked questions

Can I run Mistral locally on a Mac?

Yes — mistral-24b-4bit (Mistral Small 24B) is 12.0 GB on disk and runs comfortably on a 24 GB Mac with room for context. There's a coding-tuned Devstral in the same size class, and a 119B flagship for Studio-class machines with a lot of unified memory.

Which Mistral should I use for coding on a Mac?

Reach for devstral-24b-4bit — Devstral is the same 24B size class as Mistral Small but trained specifically for code and agent workflows. It fits a 24 GB Mac. For general chat and writing, mistral-24b-4bit is the stronger all-rounder; switch between them with one command.

Does Mistral work with Claude Code and Cursor locally?

Yes. rapid-mlx serves Mistral behind an OpenAI-compatible endpoint, so any tool with a custom base URL box — Claude Code, Cursor, Codex, Aider — can point at http://localhost:8000/v1 and use it with no bridge and no plugin.

Is running Mistral locally actually free and private?

The weights are open, rapid-mlx is Apache 2.0, and inference happens entirely on your Mac — no account, no per-token billing, and your prompts and code never leave the machine. The only costs are disk space for the weights and the electricity your Mac was already using.