Skip to main content

Download

Install an engine, open the app.

The work runs on your own machine. Each engineering app has its own small native engine; install the ones you want, then open the app in your browser and sign in there. The OpenEng AI engine is a separate download that runs open-weight models locally.

Engineering apps · freeOne small engine per app. The Terminal, Kubernetes, and Data engines are published; the Git and Chat engines are coming soon.OpenEng AI openengRuns open-weight models on your machine. Linux x86_64; needs a Console API key and a Pro plan (launching Q4 2026).

Engineering apps

One small engine per app.

Every app has its own native engine that runs on your machine and listens only on loopback (127.0.0.1). Install the engine, open the app, and sign in. The app then gives you a one-time command that starts the engine and connects the page to it.

Terminal

Learn more
Open Terminal

openeng-terminal · Linux x86_64 · macOS (Apple silicon, Intel) · Windows via WSL

Install the engine

Then open the app & sign in

terminal.openeng.app

The app gives you a one-time command to start the engine — copy it, run it, and the page connects automatically.

Kubernetes

Learn more
Open Kubernetes

openeng-kubernetes · Linux x86_64 and ARM64 · macOS (Apple silicon, Intel) · Windows via WSL

Install the engine

Then open the app & sign in

k8s.openeng.app

The app gives you a one-time command to start the engine — copy it, run it, and the page connects automatically.

Open Data

openeng-data · Linux x86_64 · macOS (Apple silicon, Intel) · Windows via WSL

Install the engine

Then open the app & sign in

data.openeng.app

The app gives you a one-time command to start the engine — copy it, run it, and the page connects automatically.

Open Git

openeng-git

Engine coming soon

The Git web app is live, but its engine hasn’t been published yet — there’s nothing to install today. The one-line installer will appear here when it ships; your OpenEng account already works across the suite.

Open Chat

openeng-chat

Engine coming soon

The Chat web app is live, but its engine hasn’t been published yet — there’s nothing to install today. The one-line installer will appear here when it ships; your OpenEng account already works across the suite.

Free: every app, every feature. Explore the apps · Every platform build

OpenEng AI

The OpenEng AI engine.

One line installs openeng, the engine that downloads, orchestrates, and runs open-weight models on your machine. The installer fetches the prebuilt binary, checks its SHA-256 when a sha256 tool is available, puts it on your PATH, and pulls the default chat model. Each config then downloads the models it references from its registry: Hugging Face, the Ollama library, ModelScope, or a direct URL.

Linux x86_64 · Windows under WSL

Before it will start

The engine boots only with an OpenEng key, an API key you create in the Console, passed as --key=<K> or the OPENENG_KEY environment variable. It registers that key with the OpenEng API, which requires an active Pro plan. Pro is launching Q4 2026, so you can install the engine today, but it starts only on a Pro account.

Status: OpenEng AI isn’t live yet. The Console and Web Console interfaces load, but the OpenEng AI API routes and the Broker aren’t deployed. Nothing can be saved or published, and no AI engine can register or connect. Pro, Enterprise, and pay-as-you-go launch in Q4 2026 and can’t be purchased yet.

See the plans

Prefer to read before you pipe to bash? Inspect install.sh. The model pull is a few GB; set OPENENG_NO_MODELS=1 to skip it. On Windows there is no native build yet: install.ps1 only prints the WSL command, wsl bash -c 'curl -fsSL https://openeng.app/engine/install.sh | bash'.

One binary

A single Rust binary, openeng: the agent loop, the embedded model runtime, and the memory system, all in one file. The same binary runs the enclosed model daemon and the headless command.

Open-weight models

Downloads open-weight models from any registry and runs them on your machine: its built-in models in its enclosed model daemon, catalog models through a local Ollama daemon you install. After the download, local inference makes no network call.

Verified

TLS download, a published SHA-256 that the installer checks when a sha256 tool is available (refusing a mismatch), and a refusal of anything that looks like a web page instead of a binary.

After the install

Start it with your Console key.

  1. 1Create an API key in the Console under Deploy → API keys. It is shown once; only its hash is stored.
  2. 2Start the engine with it: openeng serve --key=<K>, or export OPENENG_KEY and run openeng (serve is the default command). Add --config=a,b,c after serve to preload Console-published configs and download their models, and --default=<config-id> to set the fallback config.
  3. 3The engine registers the key with the OpenEng API, which checks for an active Pro plan, then dials the OpenEng Broker. It opens no network port; its model daemon uses a local Unix socket.
  4. 4Open the Web Console and start a thread. Your configs resolve by id, with their agents, models, and cells; the MCP servers the engine connects come from your .mcp.json (it doesn’t apply a config’s published MCP servers or Directories yet). If the model pull was skipped or interrupted, run openeng pull chat.

Full walkthrough in the AI docs: keys, configs, and models.

openeng [serve]
Boot the engine and dial the Broker (the default).
openeng pull <role>
Fetch an embedded model weight (chat, coding, planner, …).
openeng headless
Check your key and print the headless driver it would run; the work-pull and scheduler loops ship later.
openeng daemon
Run the enclosed model daemon (spawned by the engine).

Platforms

What's published, exactly.

Every engine build this site hosts, taken straight from the files. There is no desktop app and no standalone CLI to download: the engines are the whole install, and the apps and the Web Console run in your browser.

Published engine builds by platform
EngineLinux x86_64Linux ARM64macOS Apple siliconmacOS IntelWindows
OpenEng AI engineopenengConsole key + Pro (launching Q4 2026) Binary: openeng-linux-x86_64Not publishedNot publishedNot publishedVia WSL
Terminal engineopeneng-terminal Binary: openeng-terminal-linux-x86_64Not published Binary: openeng-terminal-darwin-aarch64 Binary: openeng-terminal-darwin-x86_64Via WSL
Kubernetes engineopeneng-kubernetes Binary: openeng-kubernetes-linux-x86_64 Binary: openeng-kubernetes-linux-aarch64 Binary: openeng-kubernetes-darwin-aarch64 Binary: openeng-kubernetes-darwin-x86_64Via WSL
Data engineopeneng-data Binary: openeng-data-linux-x86_64Not published Binary: openeng-data-darwin-aarch64 Binary: openeng-data-darwin-x86_64Via WSL
Git engineopeneng-git Coming soonNot publishedNot publishedNot publishedNot published—
Chat engineopeneng-chat Coming soonNot publishedNot publishedNot publishedNot published—
  • macOS builds are available for Terminal, Kubernetes, and Data; they are older builds pending a rebuild.
  • There is no native Windows build of any engine yet — on Windows, run the Linux engine under WSL. Every install.ps1 detects this and prints the WSL command to run instead.
  • New builds land here and on /releases first. There are no sign-up-to-be-notified forms.

Verify

Check every byte yourself.

Every installer downloads over TLS, checks the binary against its published SHA-256 when the checksum file and a sha256 tool are available, and refuses a mismatch, or a web page where a binary should be. Prefer to install by hand? Download a binary, compare its checksum, mark it executable, and move it onto your PATH.

Manual install

Download a binary from the list, run sha256sum <file> and compare the result with the value shown (it is also published next to the binary as <file>.sha256), then chmod +x it and move it onto your PATH under the engine's own name, for example openeng-terminal, or openeng for the AI engine.

Read the installers

Each one is a short script you can read before you run it: AI engine, Terminal, Kubernetes, Data. The beta engines are not code-signed yet (the macOS builds are ad-hoc signed), so the installer clears the macOS quarantine flag after the checksum step (a mismatch aborts before that).

Updates

Installed app engines check the release feed each time they start, and every 30 minutes while they run, and update themselves. They accept only a newer stable build, verify its SHA-256, then swap it in. Versions and changelogs are on /releases, read from releases.openeng.app .

Published binaries · SHA-256

Pick an app. Install its engine. Get to work.

Five free apps in your browser, each driving a small engine on your own machine (the Git and Chat engines are coming soon), and OpenEng AI for running open-weight models on your own hardware.

Explore the apps How it stays private See pricing

Installing the AI engine on servers? See headless mode and API keys.