Gemini API Computer Use is GA: screen-reading, clicking agents become a supported feature
What is it?
Computer Use lets the model look at a screenshot and return the next action, such as a click, scroll or keystroke, as a function call. Your own code (Playwright, for example) performs the action and sends back a fresh screenshot, and the loop repeats. It's now generally available in both the Interactions API and the GenerateContent API. Supported models include gemini-4-argon, gemini-3.8-flash (the one the docs recommend) and gemini-3.5-flash-lite.
pip install google-genai playwright
playwright install chromiumWhy does it matter?
With the preview label gone, it's easier to put browser automation into production: end-to-end tests for web apps, repetitive form filling, research across sites. It works in browser, mobile and desktop environments. Each action comes with an intent explaining why the model chose it, plus a safety_decision that can ask for user confirmation on risky steps. You can also opt in to prompt-injection detection that scans screenshots for hidden instructions.
Maturity
GAThe release notes list it as GA. The docs still say to run it in a sandboxed VM or container and keep a person watching for anything important.
How to track it
- Pricing: check the Gemini API pricing page to confirm Computer Use is billed like regular tokens.
- Vertex AI: see whether the Vertex AI docs reflect GA as well.
- Reference implementation: watch releases of Google's Computer Use reference repo for changes to the Docker sandbox example.