Now available for macOS & Windows

Your local AI stack, ready in minutes

Download, set up, and start using Opta CLI and Opta LMX on your machine. No API keys. No cloud. Just fast, private inference.

Get Early Access View Setup Guide

22 tok/s

on M3 Ultra

Zero

API keys needed

100%

Private, local

Scroll

Download your stack

Two tools. One local AI stack. Both free, both open.

Opta CLI

AI coding agent

The command-line interface for AI-powered coding. Ask it to build features, fix bugs, explain codebases, or write tests — all running locally on your machine.

Coming soon

Opta LMX

Local inference engine

Runs large language models on Apple Silicon via MLX. Serves a local OpenAI-compatible API endpoint so any app can use your private models — no cloud, no API key.

Coming soon

Be first to download

Drop your email and we’ll notify you the moment installers ship.

Set up in minutes

Step-by-step instructions for each tool. Pick your starting point.

Install Opta CLI

Download and run the installer for your platform, or install via Homebrew.

bash

brew install optaops/tap/opta-cli

Configure your endpoint

Point Opta CLI at your local LMX server running at port 1234.

bash

opta config set endpoint http://localhost:1234
opta config set model kimi-k2-5

Run your first command

Ask Opta CLI to fix a bug, write a test, or explain a function.

bash

opta "fix the authentication bug in src/auth.ts"

Get the most out of your stack

Tips and workflows from power users running the full Opta stack.

Performance

Use Kimi K2.5 for coding

22 tok/s on M3 Ultra

Kimi K2.5 delivers the best results for code generation, debugging, and refactoring tasks. At 22 tokens per second on M3 Ultra, it feels instant — no more waiting for API responses.

Smart Routing

Let the router decide

Smart model routing

Opta LMX includes an intelligent router that picks the right model for each request — fast models for simple queries, capable models for complex reasoning. Enable it with `opta-lmx router on`.

Productivity

Batch overnight

Queue tasks while you sleep

Got a big refactor or test suite to write? Queue it before you sleep. Opta CLI supports job queuing so your models can work through long tasks while you're away — no API rate limits, no cost per token.

More tips in the setup guide and the Opta CLI docs.

Opta Local Dashboard

Ready to manage your models?

Open the Opta Local dashboard to download models, monitor inference performance, configure routing, and manage your full local AI stack — all from a clean UI.

Open Dashboard

Requires Opta LMX running locally