Google DeepMind 推出 Co-Scientist：基於 Gemini 的多 Agent 科學研究 AI 助手★ 85
Google DeepMind Blog34 days agoReleaseGoogle DeepMind has officially launched "Co-Scientist," a new multi-agent AI collaboration system built on the Gemini model, designed to serve as a virtual…
Thinking Machines 推出原生互動模型 TML-Interaction-Small 276B-A12B：突破即時語音 SOTA 並淘汰傳統 VAD★ 85
Latent Space34 days agoReleaseAccording to AINews, the AI research team Thinking Machines (affectionately nicknamed "Team Thinky" by the community) has recently unveiled a new native…
在 AWS 上進行基礎模型訓練與推理的建構基石 (Building Blocks)★ 75
Hugging Face Blog35 days agoTutorialIn the era of generative AI, training and deploying foundation models with billions of parameters faces enormous computational and architectural challenges…
Import AI 456：遞迴自我改進與經濟成長、AI 監管的極致選擇權，以及神經電腦★ 75
Import AI (Jack Clark)35 days agoCommentaryThis issue of Import AI 456, written by Jack Clark, delves into three highly forward-looking and interconnected areas of AI development: the relationship…
Superset 如何在 Vercel 上構建 AI Agent 專屬的 IDE
Vercel Changelog36 days agoCommentaryAs AI agents rise to prominence, traditional code editors can no longer meet developers' needs for debugging, observing, and orchestrating agents. Superset is…
Anthropic 年成長達 10 倍，與其他科技巨頭裁員超過 10% 形成強烈對比★ 75
Latent Space37 days agoBusinessIn the current technology industry and broader economic environment, an extreme and striking "dual-track" divergence (dichotomy) is playing out. On one hand…
OpenAI 推出 GPT-Realtime-2、GPT-Translate 與 GPT-Whisper：全新 SOTA 即時語音 API★ 85
Latent Space38 days agoReleaseOpenAI has continued to expand the reach of its GPT-5 technology, officially launching three new voice and audio APIs: GPT-Realtime-2, GPT-Translate, and…
來自中國頂尖 AI 實驗室的內部觀察筆記★ 75
Interconnects (Nathan L.)39 days agoCommentaryThis in-depth piece from Interconnects founder Nathan Lambert documents his key observations after personally visiting several of China's top AI laboratories —…
Anthropic 與 xAI 簽署 50 億美元 Colossus 算力合約，ARR 暴增 8000%★ 90
Latent Space39 days agoBusinessThis is a heavyweight partnership that has shaken the AI industry. According to the latest report from Latent Space, AI unicorn Anthropic has reached an…
vLLM V0 到 V1 的演進：在強化學習（RL）中「正確性重於修正」的實踐★ 75
Hugging Face Blog40 days agoOpinionThis blog post published by the ServiceNow AI team delves into the major transition of the open-source large language model inference engine vLLM from V0 to…
AlphaEvolve：Google DeepMind 基於 Gemini 的程式碼 Agent 如何在各領域擴大影響力★ 80
Google DeepMind Blog40 days agoReleaseGoogle DeepMind has recently shared the latest progress and real-world impact of its new coding agent "AlphaEvolve." AlphaEvolve is an algorithmic system…
矽谷巨頭與新創轉向「AI 服務」新戰場：從賣軟體工具到直接賣勞動成果★ 80
Latent Space40 days agoCommentaryA series of recent moves from Silicon Valley and statements from major tech giants collectively point to a profound paradigm shift in the industry: AI…
Hugging Face 為 Open ASR 排行榜引入「防刷榜機制」，使用私有測試數據打擊 Benchmaxxer★ 75
Hugging Face Blog41 days agoReleaseHugging Face has recently made a major update to its popular Open ASR (Automatic Speech Recognition) leaderboard, aimed at combating the increasingly serious…
OpenAI 物理學家 Alex Lupsasca 專訪：GPT-5.x 如何在理論物理與量子重力領域推導出全新研究成果 (Doing Vibe Physics)★ 85
Latent Space41 days agoCommentaryThis interview records an in-depth conversation between OpenAI theoretical physicist Alex Lupsasca and Latent Space, centered on how GPT-5.x — OpenAI's…
[AINews] 工具還是「他者」？從 Clippy 與 Anton 之爭探討 AI 的角色本質
Latent Space42 days agoCommentaryIn today's era of rapid AI iteration, we often focus on model parameters and benchmarks while overlooking the most fundamental question of product design…
蒸餾恐慌：為什麼將「知識蒸餾」稱為安全攻擊是極其糟糕的趨勢★ 75
Interconnects (Nathan L.)42 days agoOpinionIn the field of machine learning, "knowledge distillation" is a well-established technique that generally refers to using the output data generated by a…
Import AI 455：AI 系統即將開始自我構建——邁向遞迴自我提升的第一步★ 85
Import AI (Jack Clark)42 days agoCommentaryIn the latest issue of Import AI 455, Jack Clark guides readers through an exploration of a highly forward-looking and both exciting and concerning theme: AI…
Google DeepMind 探索醫療新模式：研發「AI 協同臨床醫生」推動 AI 輔助醫療★ 75
Google DeepMind Blog46 days agoReleaseGoogle DeepMind has recently published new research and a vision for an "AI co-clinician." As healthcare systems worldwide face unprecedented pressure —…
IBM 推出 Granite 4.1 開源大語言模型：架構設計與構建技術深度解析★ 75
Hugging Face Blog47 days agoReleaseIBM has published a detailed blog post on Hugging Face outlining the construction technology and architectural design behind its latest generation of…
DeepInfra 正式加入 Hugging Face 推理服務商（Inference Providers）陣容 🔥★ 72
Hugging Face Blog48 days agoReleaseHugging Face's official blog has announced that DeepInfra — a well-known high-performance, low-cost serverless inference platform — has officially joined…
NVIDIA 推出 Nemotron 3 Nano Omni：支援長文本的多模態智慧模型，專為文件、語音與影片 Agent 設計★ 75
Hugging Face Blog48 days agoReleaseNVIDIA has officially launched a new lightweight multimodal model, "Nemotron 3 Nano Omni." This model is designed to deliver powerful multimodal intelligence…
Google DeepMind 與大韓民國宣布建立合作關係，利用前沿 AI 模型加速科學突破★ 75
Google DeepMind Blog49 days agoBusinessGoogle DeepMind officially announced on April 27, 2026, a strategic partnership with the government of the Republic of Korea (South Korea). The core objective…
如何使用 OpenAI 的 Privacy Filter 打造具備高擴展性的 Web 應用程式★ 75
Hugging Face Blog50 days agoTutorialIn the current era of booming generative AI, one of the greatest challenges enterprises and developers face when adopting large language models (LLMs) is "data…
DeepSeek-V4：Agent 真正能派上用場的百萬 Token 超長上下文★ 85
Hugging Face Blog53 days agoReleaseDeepSeek has introduced its next-generation open-source model, DeepSeek-V4, whose most attention-grabbing feature is an ultra-long context window of up to 1…
未來的預兆：GPT-5.5 與 AI 指數型成長的下一步★ 85
One Useful Thing (Mollick)53 days agoCommentaryWharton School professor Ethan Mollick, writing in his well-known newsletter "One Useful Thing," has published a profound analysis of GPT-5.5. He describes…
Decoupled DiLoCo：Google DeepMind 推出更具彈性的分散式 AI 訓練新技術★ 85
Google DeepMind Blog54 days agoReleaseGoogle DeepMind has recently unveiled a new distributed AI training technique called "Decoupled DiLoCo." This technology represents a major upgrade to its…
QIMMA ⛰：首個品質優先的阿拉伯語大型語言模型（LLM）排行榜
Hugging Face Blog55 days agoReleaseThe Technology Innovation Institute (TII) of the United Arab Emirates — the organization behind the well-known open-source model Falcon — has officially…
AI 與網路安全的未來：為什麼「開放」至關重要★ 75
Hugging Face Blog56 days agoOpinionAs artificial intelligence (AI) technology undergoes explosive growth, cybersecurity has become a focal point of concern for governments and enterprises…
解讀當前開源與閉源 AI 模型的性能差距：超越單一評估指標的迷思★ 75
Interconnects (Nathan L.)56 days agoOpinionIn today's AI landscape, the performance gap between open-weights models (such as Meta's Llama family) and closed-source models (such as OpenAI's GPT and…
Import AI 454：自動化對齊研究、中國 AI 模型安全評估與全新 4 位元浮點格式 HiFloat4★ 75
Import AI (Jack Clark)56 days agoCommentaryIn this issue of Import AI 454, written by Jack Clark, the author begins by posing a thought-provoking question about finance and sociology: "At what point…