Overview
Gemini 2.5 Computer Use is a specialized preview model built on Gemini 2.5 Pro’s visual understanding and reasoning capabilities. It was announced on October 7, 2025 for building agents that interact with graphical user interfaces. (blog.google)
Capabilities and uses
The model interprets screenshots and proposes actions such as clicking, typing, scrolling and navigating. Applications must provide the execution environment, carry out the proposed actions and return results to the model. Intended uses include browser automation, form filling, information gathering, multi-step software workflows and UI testing. (aistudio.google.com)
The model card lists text and image input, text output, a 128,000-token input limit and a 64,000-token output limit. (aistudio.google.com)
Limitations and availability
This is a legacy preview optimized for browser-based computer use. Agents can make mistakes and face risks including prompt injection, malicious web content and unintended high-impact actions. Developers should test their systems, use safety controls and require user confirmation where appropriate. (ai.google.dev)
Google’s model-specific documentation states that the Gemini API model was shut down on July 28, 2026. Although some general Google documentation still lists the model, that listing should not be treated as proof of current availability. Google recommends migrating to Gemini 3.8 Flash or another Gemini 3 model with computer-use support. (aistudio.google.com)
Pricing
Checked November 17, 2026. Current Gemini Developer API input and output pricing for this exact model could not be verified on the current official pricing page. The model-specific Gemini API documentation reports that it has been shut down. (aistudio.google.com)
Google Cloud’s official Agent Platform pricing table still lists Gemini 2.5 Pro Computer Use-Preview under standard pricing, in USD per 1 million tokens: for prompts of at most 200,000 input tokens, $1.25 input and $10.00 text output, including response and reasoning tokens. For prompts above 200,000 input tokens, the table lists $2.50 input and $15.00 output. The latter pricing tier does not establish that this model supports such prompts; its Gemini API model card lists a 128,000-token input limit. The table lists caching, priority, and Flex/Batch pricing as unavailable for this entry. A pricing-table entry alone does not establish current endpoint availability. (cloud.google.com)