谷歌DeepMind發布Gemini 4 Carbon，在多個指標上取得進步
谷歌deepmind發布gemini 4 carbon，其最新前沿模型，在人工分析智能指數v5上獲得59分，超過argon的50分。
google deepmind released gemini 4 carbon, its latest frontier model, two weeks after argon. carbon achieved 59 points on the artificial analysis intelligence index v5, up from argon's 50 points. it is designed for multi-day, autonomous development workflows and features advanced cybersecurity capabilities, currently undergoing the u.s. government's voluntary pre-testing process. carbon leads the overall aa index but trails claude opus 5.5 in agentic coding. google reports improvements in scientific reasoning, multimodal understanding, and full-stack web development.