谷歌加入CUA战场发布Gemini 2.5 Computer Use：AI直接操作浏览器

谷歌的 Computer Use 模型来了！

 

今天凌晨，谷歌 DeepMind 重磅发布了基于 Gemini 2.5 的计算机使用模型Gemini 2.5 Computer Use。

 

考虑到前些天谷歌才刚刚发布了 Chrome DevTools (MCP)，Gemini 2.5 Computer Use 的诞生倒不是特别让人惊讶。简单来说，与 OpenAI 的 Computer-Using Agent (CUA) 类似，DeepMind 的这个模型可让 AI 直接控制用户的浏览器 —— 在视觉理解和推理能力的基础上，该模型可以帮助用户在浏览器中执行点击、滚动和输入等操作。

 

 

 

 

先来看两个官方演示。

 

提示词：From https://tinyurl.com/pet-care-signup , get all details for any pet with a California residency and add them as a guest in my spa CRM at https://pet-luxe-spa.web.app/. Then, set up a follow up visit appointment with the specialist Anima Lavar for October 10th anytime after 8am. The reason for the visit is the same as their requested treatment.

 

 

提示词：My art club brainstormed tasks ahead of our fair. The board is chaotic and I need your help organizing the tasks into some categories I created. Go to sticky-note-jam.web.app and ensure notes are clearly in the right sections. Drag them there if not.

 

 

可以看到，不管是收集网络信息与执行动作，还是整理杂乱笔记，Gemini 2.5 Computer Use 都非常准确地完成了任务，同时速度也相当快。

 

在相关基准上，Gemini 2.5 Computer Use 的性能表现也达到了 SOTA 水平：