AI 導讀
10 月 27 日 AI 產業每日摘要涵蓋五大新聞。
- Genie 4:Google DeepMind 於 10 月 27 日發布的多模態模型,能即時構建具備聲音與語音互動的虛擬世界。
推薦理由
Genie 4 的即時世界建構與語音互動,展示了多模態模型在即時生成上的突破。
thehype.@thehypedotnews
·5 小時前
原文內容 (English)
october 27 in ai, in five stories from our daily digest
genie 4 from google deepmind builds worlds that now have their own sound and talking characters, in real time.
apple is putting an ai assistant into pages, keynote and numbers – plus mcp, so outside agents can edit documents.
machenta raised $40m for a robot model that learns a task from one demo: 74 of 74 block placements in its trials.
oracle will sell google tpus, cerebras chips and nvidia vera rubin in one cloud.
and the rest: lfm3 open models at ~2b, a 43.1m-line lean proof, $30m for gpu resale
full article with more news in the comments
推文來源 · X (Twitter)在 X 參與討論