谷歌推出Gemma 4开源模型及第八代TPU
谷歌发布了gemma 4，称其为最强大的开源模型，并在cloud next '26大会上推出了第八代tpu。公司还推出了用于业务工作流的gemini enterprise agent平台和用于数据分析的deep research max。在32000名参会者面前，谷歌发布了260多项公告，ceo桑达尔·皮查伊指出75%的云客户现已使用谷歌云ai。
Google detailed its April 2026 AI advancements, headlined by the release of Gemma 4, which the company claims is byte-for-byte the most capable open model. Cloud Next '26 served as the backdrop for the launch of the Gemini Enterprise Agent Platform, designed for building and governing autonomous agents, and new eighth-generation TPUs built for the compute demands of agentic workflows. Google CEO Sundar Pichai revealed that 75% of Google Cloud customers now utilize the company's AI tools, with 330 organizations processing over a trillion tokens in the past year. The update also introduced Deep Research Max for advanced analysis and Google Vids, a free professional video creation suite.