The new family of AI models can run on a smartphone, a Raspberry Pi, or a data centre, and is free to use commercially.

A University of Arizona fraternity has been suspended from campus after an investigation by its national headquarters, officials said Friday. Sigma Alpha Epsilon headquarters notified UA officials ...

AMD adds Day 0 support for Google Gemma 4 across Radeon, Instinct, and Ryzen AI, enabling full-stack AI deployment.

Google has launched Gemma 4, which goes beyond chatbots and creates AI agents that can plan tasks, take actions on their own, generate code even without internet access, and process audio and video.

腾讯前沿科技论文解读专栏，在代码与商业的交汇处，寻找AI的确定性。文｜博阳编辑｜徐青阳随着任务的复杂度提升，Agent（智能体）的上下文在无限膨胀。在无穷的历史对话、工具调用输出、中间步骤以及报错信息中，模型迷糊了，于是开始跳步、忽视、绕道。这就是一 ...

今年 4 月初（也就是前几天），Google 最新一代开源模型 Gemma 4 家族正式发布，上线当天就冲上 Arena AI ...

凌晨，硅谷又炸了锅。 谷歌DeepMind毫无征兆地甩出Gemma 4全家桶——四款开源模型，从2B到31B，全部基于Gemini 3同源技术打造。这不是一次常规更新，而是对开源大模型格局的彻底重写。 最让同行后背发凉的数字：31B Dense模型，Elo评分1452，开源榜单第三。排它前面的两位，参数分别是它的20倍和30倍。用三十分之一的体量，打出了同等水平的成绩——参数效率这个词，被谷歌重新 ...

自从第一代 Gemma 在 2024 年 2 月发布以来，开发者们已经下载了超过 4 亿次，社区创造了超过 10 万个变体。谷歌把这叫做“Gemmaverse”——一个围绕 Gemma 建立起来的生态宇宙。

Google dropped Gemma 4 on April 2, 2026, and it's a game-changer for anyone building AI. These open models pull smarts straight from Gemini 3, Google's top ...