谷歌发布最新大模型,各指标全面登顶,AI规模化应用再进一步
Xuan Gu Bao·2025-11-19 00:35

Group 1 - Google officially launched its most powerful AI model, Gemini 3, which is now available on Google Search, the Gemini app, and various developer platforms [1] - Gemini 3 achieved a historic high score of 1501 on the global AI model LMArena leaderboard and surpassed the previous record of 31.64% in the Humanity's Last Exam benchmark with a score of 37.5% [1] - The model is recognized for its advanced reasoning capabilities and is described as the best multimodal understanding model and the safest model from Google [1] Group 2 - Google introduced a developer platform called Google Antigravity, which leverages Gemini 3's advanced reasoning and tool usage capabilities, integrating the Gemini 2.5 Computer Use model and the Nano Banana image editing model [2] - The AI industry is undergoing a critical transition from technological breakthroughs to large-scale applications, with improvements in model performance and decreasing usage costs accelerating industry penetration [2] - New Asia Electronics has entered the supply chain of major server manufacturers such as Dell, HP, Google, Amazon, and Microsoft with its high-frequency, high-speed copper cable connectors [2][3]