Anthropic Gives Claude 3.5 Sonnet Actual Desktop Control
The new Computer Use API lets Claude 3.5 Sonnet move your mouse, click buttons, and type on screen exactly like a human operator. Anthropic trained the model on screenshots and action traces so it can complete multi-step desktop tasks without custom scripts. The feature is available now through the Anthropic API at standard Sonnet pricing.
You stop treating the model as a text box and start giving it the same interface a human employee would use. This shifts your workflow from prompt engineering to task decomposition and verification loops. Solo builders can now replace brittle scripts or expensive RPA tools with a single general-purpose agent.
Anthropic documented the release at anthropic.com/news/3-5-models-and-computer-use. Early internal tests showed the model completing 20-30 percent of realistic office tasks such as spreadsheet updates and form submissions without further human input.
Step 1: Sign up for Anthropic API access and enable the computer-use beta flag in your account settings. Step 2: Send a request that includes both a goal description and a live desktop screenshot using the computer_use_20241022 tool. Step 3: Review the returned action list, approve or correct the first step, then iterate until the task completes.