OpenComputer

Free plan APILinuxMacWindows

What OpenComputer is

OpenComputer is an open-source desktop agent for operating applications through what it sees on screen and mouse and keyboard actions. It can click, drag, scroll, enter text, use keyboard shortcuts, open URLs, and wait for changes in the interface. For its vision-language model, it can connect to a local model through vLLM or to a compatible hosted API. The chosen endpoint must accept image inputs and function tools and support the model’s tool-call format.

The default Ask first mode requests approval before mouse, keyboard, and URL actions; Automatic mode performs those actions without individual prompts. OpenComputer currently works only with the primary display. Linux X11 is its best-supported environment, while Wayland, Windows, and macOS are experimental. Build targets include Linux AppImage and deb packages, a Windows NSIS installer, and a macOS DMG.

The project says screenshots, prompts, and tool results can stay on the user’s machine. API keys are stored in the per-user application-data directory with user-only file permissions, but not in the operating-system keychain. OpenComputer is in early alpha, and its maintainers advise supervising desktop actions, especially in sensitive or irreversible workflows. It is licensed under MIT; third-party dependencies and model weights have their own licenses.

Who it suits

OpenComputer suits people who want a local-first agent to operate desktop applications and can provide a compatible vision-language model endpoint. It is a better fit for users prepared to supervise an early-alpha tool, particularly on Linux X11, than for workflows requiring multi-monitor support or a fully established platform experience.

What it does well

Controls apps with mouse, keyboard, URL, and scrolling actions.
Can use a local vLLM model or compatible hosted API.
Ask first mode requires approval for desktop actions.
Screenshots, prompts, and tool results can stay on the user's machine.
Open source under the MIT License.

What to know first

Only the primary display is supported currently.
Wayland, Windows, and macOS are experimental.
API keys are not stored in the operating-system keychain.
Automatic mode can act with the user's desktop privileges.

Limits

The agent currently operates only on the primary display. Linux X11 is best supported; Wayland, Windows, and macOS are experimental. The project is early alpha and recommends supervising desktop actions, especially for sensitive or irreversible workflows.

Getting started

Open the OpenComputer project on GitHub.
Choose a build target for Linux, Windows, or macOS.
Provide a local vLLM model or compatible hosted API endpoint.
Confirm the endpoint supports image inputs, function tools, and the model’s tool-call format.
Use Ask first for approval before mouse, keyboard, and URL actions.

Verdict

Choose OpenComputer if you want an open-source desktop agent with screen understanding, configurable model options, and approval prompts for actions. Look elsewhere if you need multi-monitor automation or a mature, established experience on Wayland, Windows, or macOS.

Questions

Is OpenComputer free?

Its pricing model is free, and the project is open source.

Which operating systems does it support?

Build targets are Linux AppImage and deb, Windows NSIS, and macOS DMG. Linux X11 is best supported; Wayland, Windows, and macOS are experimental.

Can it use a local model?

Yes. It can use a local vision-language model through vLLM or a compatible hosted API. The endpoint must support image inputs, function tools, and the model’s tool-call format.

Does it ask before taking desktop actions?

Ask first is the default and requires approval for mouse, keyboard, and URL actions. Automatic mode performs actions without individual prompts.

Is OpenComputer open source, and what is its license?

Yes. OpenComputer uses the MIT License; third-party dependencies and model weights retain their own licenses.

Who makes OpenComputer?

The linked GitHub profile identifies the maker as Andrew Krikorian and lists California.

Facts from OpenComputer's own pages

Action review: Ask first is the default permission mode and requires approval for mouse, keyboard, and URL actions; Automatic mode runs actions without individual prompts. · github.com, 7 Oct 2026
API compatibility: The configured endpoint must support image inputs, function tools, and the model’s tool-call format. · github.com, 7 Oct 2026
Approvals: Ask first is the default permission mode and requires approval for mouse, keyboard, and URL actions. · github.com, 7 Oct 2026
Credential storage: API keys are stored in the per-user application-data directory with user-only file permissions, but are not yet backed by the operating-system keychain. · github.com, 7 Oct 2026
Desktop controls: It supports clicking, dragging, scrolling, text entry, keyboard shortcuts, opening URLs, and waiting for UI changes. · github.com, 7 Oct 2026
Display limit: The current agent operates only the primary display, and multi-monitor automation is on the roadmap. · github.com, 7 Oct 2026
Installer formats: Build targets are Linux AppImage and deb, Windows NSIS, and macOS DMG. · github.com, 7 Oct 2026
Integrations: The current model boundary uses an OpenAI-compatible Chat Completions API; dedicated Discord, search, and MCP integrations are listed as future work. · github.com, 7 Oct 2026
License: OpenComputer is licensed under the MIT License, while third-party dependencies and model weights retain their own licenses. · github.com, 7 Oct 2026
Maker: The linked GitHub profile identifies the maker as Andrew Krikorian and lists California. · github.com, 7 Oct 2026
Model compatibility: The model endpoint must support image inputs, function tools, and the model's tool-call format. · github.com, 7 Oct 2026
Model options: It can use a local vision-language model through vLLM or a compatible hosted API. · github.com, 7 Oct 2026
Privacy and security: The README says screenshots, prompts, and tool results can stay on the user's machine, and that credentials, screenshots, model calls, and automation remain in Electron's main process. · github.com, 7 Oct 2026
Project maturity: The project describes itself as early alpha and warns that desktop actions should be supervised, especially around sensitive or irreversible workflows. · github.com, 7 Oct 2026
Purpose: OpenComputer is an open-source, local-first desktop agent that understands the screen and operates applications through mouse and keyboard interactions. · github.com, 7 Oct 2026
Security design: The Electron renderer is sandboxed with Node.js integration disabled, while privileged model and desktop automation behavior runs in the main process. · github.com, 7 Oct 2026
Security limitations: The project says its controls are defense in depth rather than a complete sandbox, and Automatic mode lets the model act with the user's desktop privileges. · github.com, 7 Oct 2026
Security reporting: The security policy asks users to report vulnerabilities privately and says reports should receive an initial response within seven days, though response times may vary because it is a volunteer project. · github.com, 7 Oct 2026
Supported systems: Build targets are Linux AppImage/deb, Windows NSIS, and macOS DMG; Linux X11 is best supported, while Wayland, Windows, and macOS are experimental. · github.com, 7 Oct 2026
What it does: OpenComputer is an open-source desktop agent that understands the screen and can operate applications with mouse and keyboard interactions. · github.com, 7 Oct 2026

Where OpenComputer stands on Freedom251

AI Computer Use Agents#10 of 24