Self-Operating Computer vs UFO
Which should you use? A side-by-side look at what each one does, who it suits, what it costs, how hard it is to set up, and where each falls short.
Open source · MIT
Lets an AI model look at your screen and control the mouse and keyboard to complete a task.
Open source · MIT
Open-source Windows agent that uses an LLM to operate desktop apps, and a multi-device orchestration layer cal
The short answer
Choose Self-Operating Computer if you need…
- Developers exploring AI computer-use agents
- Researchers comparing vision-capable models on desktop tasks
- Students learning how screen-reading agents work
- People who already have an OpenAI or other supported model API key
Choose UFO if you need…
- Developers experimenting with LLM-driven Windows desktop automation
- Researchers studying multi-agent task orchestration
- People who want to automate repetitive tasks in Windows apps and are comfortable with setup
- Teams exploring cross-device workflows, using the Galaxy framework
Side-by-side
| Self-Operating Computer | UFO | |
|---|---|---|
| Type | Framework | AI model / agent |
| Price | Free and open source | Free and open source |
| Licence | MIT | MIT |
| GitHub stars | 10.3k | 10.2k |
| Setup difficulty | Intermediate | Advanced |
| How you get it | Install with a package manager | Clone and build from source |
| Main language | Python | Python |
| Last code update | 17 days ago | 6 days ago |
What each one is
Self-Operating Computer
This is a Python framework that lets a multimodal AI model view your computer screen and decide on a sequence of mouse and keyboard actions to reach a goal you give it. The README says it uses the same inputs and outputs a human operator would. It is commonly used to experiment with AI 'computer use' and to compare how different vision models perform at operating a computer.
UFO
UFO³ is a Microsoft research project for building AI agents that control Windows computers and coordinate work across several devices. Its older part, UFO², runs on a single Windows machine and uses a language model to read the screen and act on apps through clicks and API calls. The newer Galaxy part splits a request into a graph of tasks and assigns them to different devices, such as Windows, Linux, or Android, as described in the README.
Best for
Self-Operating Computer
- Developers exploring AI computer-use agents
- Researchers comparing vision-capable models on desktop tasks
- Students learning how screen-reading agents work
- People who already have an OpenAI or other supported model API key
UFO
- Developers experimenting with LLM-driven Windows desktop automation
- Researchers studying multi-agent task orchestration
- People who want to automate repetitive tasks in Windows apps and are comfortable with setup
- Teams exploring cross-device workflows, using the Galaxy framework
Strengths
Self-Operating Computer
- Supports several multimodal models, including local options via Ollama according to the README
- Installs with a single pip command and runs with a single command
- MIT licensed and open source with an active community Discord
UFO
- Open source under the MIT license with a public paper and documentation site
- Two tiers: UFO² is described as simpler and stable, while Galaxy adds multi-device orchestration
- Combines GUI actions with API calls, which the README says can reduce LLM calls
Not ideal for
Self-Operating Computer
- Anyone who needs reliable unattended automation, since the README warns that error rates are high for some models such as LLaVA and that it may not work as expected during model outages
- Users who do not want a model reading their screen and sending content to a third-party API
- Windows or Linux users who cannot grant screen and input access, since the README requires those permissions
UFO
- Non-technical users wanting a ready-made consumer app, since the README describes setup and configuration steps
- Anyone needing a stable, fully tested production system, because the README labels Galaxy as in active development
- Users who need macOS-only automation, since the README lists Windows, Linux, and Android as supported device types
Trade-offs
Self-Operating Computer
- The README states error rates are very high with LLaVA and the project is described as a base to build on
- Depends on paid third-party model APIs for most options, and gpt-4o access requires a minimum spend noted in the README
- Requires granting screen recording and accessibility permissions on macOS
UFO
- Depends on a language model and its API access, which the README discusses but this summary does not detail
- Galaxy setup is described as moderate in difficulty, and it is still in active development
- Primary focus is Windows, so cross-platform reach depends on the Galaxy device options
Getting started
Self-Operating Computer
Install the project
pip install self-operating-computerRun the project
operateUFO
Clone the repository
git clone https://github.com/microsoft/UFO.gitFrequently asked questions
What is the difference between Self-Operating Computer and UFO?
Self-Operating Computer: Lets an AI model look at your screen and control the mouse and keyboard to complete a task. UFO: Open-source Windows agent that uses an LLM to operate desktop apps, and a multi-device orchestration layer cal. Self-Operating Computer suits developers exploring ai computer-use agents, while UFO suits developers experimenting with llm-driven windows desktop automation.
Are Self-Operating Computer and UFO free?
Self-Operating Computer: Free and open source. UFO: Free and open source.
Which is easier to set up, Self-Operating Computer or UFO?
Self-Operating Computer is easier to get started with (rated intermediate); UFO is rated advanced and usually needs more setup.
Is Self-Operating Computer or UFO more popular?
Self-Operating Computer has more GitHub stars (10.3k) than UFO (10.2k). Popularity isn't the same as fit — check which one matches your use case above.
What are other alternatives to Self-Operating Computer and UFO?
Other options people compare them with include wshobson/agents, RD-Agent, DeerFlow, SWE-agent, Open SWE.
Keep comparing
Built from our catalogue entries for both products: open-source details come from each project's README (AI-assisted, refreshed weekly) and GitHub data; AI-tool details are editorial. Spotted something wrong? Let us know from either product's page.