Pai 的 AI 雷达
2026-08-28：[AINews] OpenAI to reach AGI；Breaking Claude Code Opus 5 ；Warp 如何让 Agent 自我进化
Contact-rich manipulation requires adapting to contact states that can evolve substantially within an action horizon. However, chunk-based vision-language-action models predict complete action chunks from observations co
Explicit visual intermediates can help multimodal large language models (MLLMs) externalize spatial evidence and updated visual states, but their utility depends on whether an image editor can faithfully realize the requ
Reusable skill libraries allow large language model (LLM) agents to reuse procedural knowledge across tasks, but they also turn memory access into a challenging retrieval problem. Full-library prompting preserves coverag
Conventional video editing primarily focuses on scene-level content, whereas live streaming places greater emphasis on the human subject. However, directly applying existing video-editing methods to human-centric live st
Evaluation artifacts specify a forward computation: a task, scorer, and reported metric. They do not necessarily license the claim attached to that metric because the historical evidence and alternative semantics needed
Book hotels and track airfares, plus view miles and rewards with AI Mode in Google Search.
Use Claude, Anthropic's AI assistant, right from your terminal. Claude can understand your codebase, edit files, run terminal commands, and handle entire workflows for you.
Midjourney Updates 宣布 V8.2 编辑模型进入测试，强调在现有图上改局部、提画质，而不是只重新文生图。
技术媒体拆解 Microduck：开源、强化学习训练步态、消费级价格，面向开发者与爱好者。
GLM-5.3 现已开放权重。 我们最强大的智能体编码与网络防御模型，现已可供下载、运行和定制。 权重：https://huggingface.co/zai-org/GLM-5.3 技术博客：https://z.ai/blog/glm-5.3 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmtd32q
斯坦福大学研究人员领衔发布 Terminal-Bench-Science 0.1，用来自生命、物理、地球、数学和工程科学的 70 个专家精选任务评估 AI 智能体的科研能力。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmtcxdp3f07l3roq5uk6om1aw
斯坦福大学研究人员领衔发布 Terminal-Bench-Science 0.1，用来自生命、物理、地球、数学和工程科学的 70 个专家精选任务评估 AI 智能体的科研能力。
IEEE Spectrum 把 Hugging Face / Pollen 的 399 美元开源双足 Microduck 做成 Video Friday 主角。
BBC 回顾机器人披萨制作的失败案例，说明接触物理、卫生与成本让餐饮自动化反复翻车。
媒体称 Google 新 Flash 视频模型在价格和镜头控制上同时推进，适合更长镜头与后期放大。
404 Media 报道美国 ICE 拟花费最多约 200 万美元采购 Boston Dynamics 机器狗，用于执法现场。