AI 情报文章归档

Sierra 围绕智能体 Ghostwriter 重新设计 Agent Studio，用户可在同一会话中排查解决率下降原因、查看对话、修改 journey 并测试修复。

Squarespace 宣布重组并裁员约19%，涉及约300名员工，公司发言人称此举旨在简化运营、聚焦优先事项。文章指出 vibe-coding 建站工具正在颠覆行业，并提到 Wix 以 8000 万美元收购 Base44、Lovable 8 月完成 4 亿美元 C 轮融资估值 133 亿美元。

纽约时报在曼哈顿联邦法院起诉隐私浏览器公司 Brave，指控其抓取该报版权内容并分发给 Anthropic、Perplexity、Mistral、xAI 和 Cohere 等客户；同日 Daily News、Chicago Tribune 和 Orlando Sentinel 提起类似诉讼。

根据我们对 26K+ 会话的分析，Agent Arena 中排名前十的智能体每生成一个 token，大约要处理 93 个输入 token。 这就是各家实验室竞相降低缓存输入成本的原因。 下方视频展示了缓存在实践中的运作方式，用的是来自 @AnthropicAI 的 Opus 5.5 和 @GoogleDeepMind 的 Gemini 4 Argon 的真实会话。 深入了解这场隐藏的 AI 价格战：@DawidGalarowicz 的文章见下方。

我们最近在 Claude Code 中集成了消息线程。在 500K 到 1M token 的请求上，首 token 中位时间降低约 35%。 在长对话中，请求中位数从约 800 KB 降至 1.8 KB。

Anthropic 的 Messages API 推出 Message threads 公开测试版。API 会在服务端缓存轮次之间的对话，开发者只需发送新消息和上一条消息 ID，无需重发完整对话，Claude 可更早开始响应。

A prompt-injection benchmark, three bugs in my own scorer, and the result I only found after fixing...

Agents burn turns guessing how to test, lint and reset your DB. Ten make targets, two lines in AGENTS.md, and make check as the done check.

Sensorized human demonstrations offer scalable data for mobile manipulation, but embodiment differences complicate policy learning. We present Pixel-UMI, a pipeline for learning environment-aware whole-body policies from human data. We fir…

General-purpose mobile manipulation requires a combination of capable robot hardware, scalable demonstration collection, expressive whole-body control, and effective policy learning. Yet these components are often developed independently, …

Chain-of-Thought (CoT) generation has substantially improved the performance of Large Language Models (LLMs) on complex reasoning tasks, including code generation, data analysis, and exam-style question answering. Despite these advances, p…

GUI web agents require realistic and interactive environments that remain stable during both evaluation and training. Live website evaluation suffers from changing content, access restrictions, and limited coverage of authenticated workflo…

Muon has recently emerged as a promising optimizer for pretraining large language models, offering a speedup over AdamW in minimizing pretraining loss. However, it remains unclear whether Muon's advantage persists after fine-tuning. In thi…

Anthropic 更新了 Claude Code 在 Opus 5.5 上的系统提示词，新增两句指令，引导模型不要复刻特定现有游戏，避免 IP 侵权。Anthropic 的 Addy Osmani 回应称原创游戏、类型游戏和用户自设资产的游戏仍可正常制作，公司会继续迭代寻找平衡；此前复古游戏圈出现大量 AI 制作的破解移植作品。

Reinforcement learning with verifiable rewards (RLVR) can enable generalizable improvement for large language model (LLM) reasoning. This paper studies how to incentivize such generalization via reasoning strategies. We study this question…

Group-based reinforcement learning with verifiable rewards (RLVR) learns only from prompts whose sampled completions disagree: all-pass and all-fail groups produce zero relative advantage and burn generation budget without moving the polic…

Communication is important for our well-being, but it can be impaired by disabilities, such as locked-in syndrome (LIS), cerebral palsy (CP), and dysarthria. Non-invasive brain-to-text brain-computer interfaces (BCIs) offer a promising sol…

Vision-Language Models (VLMs) hold promise for zero-shot video task-progress estimation in robot learning, yet existing frameworks frequently suffer from confirmation bias, hallucinating progress in failed executions due to implicit succes…

A model is only half of an AI agent system. The other half is the inference strategy: the reasoning procedure that turns model calls into a response. Today that second half is written by hand, and the experience of building one strategy ne…

Large language models (LLMs) show growing promise for scientific discovery, yet basic questions remain about what enables reliable discovery. How can we trace the scientific discovery capacity of LLMs to their training data? We study this …

Evaluating a generative model on all queries from all available benchmark tasks is prohibitively expensive. As such, models are often evaluated on different subsets of tasks and/or different subsets of queries from a given task. The result…

Newcomer 报道多家创业公司尝试把 GPU 算力变成可交易的大宗商品。

Increased computing power has transformed computational science and improved predictive capabilities for complex systems, but many phenomena still exceed current simulation capabilities or require rapid evaluations for fast design iteratio…

对话越长，语音智能体的成功率越高。 我们分析了 469 个在生产环境中同时运行语音和聊天的智能体，发现了一个有趣的规律。当对话变得复杂时，语音是解决问题最快的路径。 https://elevenlabs.io/blog/voice-wins-complex-stressful-conversations

Anthropic 更新的 S1/IPO 申报文件显示未来模型可能很快达到其递归自我改进（RSI）阈值，即 AI 能力以无重大 AI 帮助时最快持续速度的 2 倍提升，1 年相当于 2 年进展。

ufakzeka-kulak is an open Turkish speech-to-text model of 31.6M parameters, a FastConformer encoder with a token-and-duration transducer (TDT) and a CTC head, trained from scratch and released under Apache-2.0. It writes cased, punctuated …

Deep Principle 发布基于 GLM-5.3 后训练的开放权重科学模型 Hydrogen-1.0，753B 参数（MoE，约 40B 激活），1M token 上下文，MIT 许可证，权重已在 Hugging Face 和 ModelScope 上线。

As pretraining scales, medical language models have advanced rapidly in clinical question answering. Their ability to achieve effective physiological control through sustained intervention still lacks systematic quantitative evaluation. Ty…

AI agents are increasingly deployed in settings where they interact and coordinate with other agents, yet honesty is mostly measured in single-turn, single-agent benchmarks that pressure the model to deceive. We study deception that emerge…

In this paper, we study when two parties should pool their data to train a machine learning model. Two agents, Alice and Bob, each hold a sensitive labeled dataset and want to train a classifier from a given hypothesis class $\mathcal{H}$.…

Understanding how neurons respond to visual stimulus is an important question in both deep learning and neuroscience. It has significant implications for enhancing the interpretability of black-box artificial neural networks and for unders…

We show that the causal structure of spacetime can emerge within the memory of a learning agent performing a quantum information task. We train a model-based reinforcement-learning agent to maximize the quantum mutual information between l…

Neural operators provide efficient surrogates for computational fluid dynamics (CFD), but most are designed for static geometries and are not equipped to model unsteady flows induced by continuously moving boundaries. We study dynamic-boun…

苹果计划最早 2027 年推出代号 J595 的家用机器人，可能命名为 HomeView Ultra，售价或达 1000 美元以上，开发延期则可能推至 2028 年。该 AI 驱动机器人可自主移动、支持 FaceTime 通话，并在有人进入房间时自动开灯。报道称其机器人团队仅约 10 名核心工程师，软件团队约 300 人。

Skyfall AI 发布 SKY-001，可按零件与建造步骤在 3D 中设计实体产品，CAD 精度与 GPT-6 Astra 相当，单次设计成本约低 150 倍。模型输出零件树、每步操作草图及中间实体，由 Hierarchical Composer 拆解装配与零件、Construction World Model 预测建造视觉步骤两阶段完成。

How I wrapped llama.cpp in a small, coroutine-first Kotlin API, and the three problems that took the most time: threads, cancellation and broken emoji.

据 Telegraph 报道，SpaceX 上周末削减了 Grok Bot 的用量配额，多名用户配额减半。Musk 在 X 上回应称运行 Grok Bot 的成本相对收入高到难以持续，但未直接确认削减。

OpenAI 发布了 LifeVal，一个测试 AI 模型能否替人们完成日常事务的基准。 GPT-6 Astra 在日常任务上得分 80.6%，而不到 9 个月前 GPT-5.2 的得分为 49.3%。 LifeVal 的任务涵盖照护、健康、教育、金融、购物和家庭行政事务，从预约车辆年检到在预算内订购清真食品杂货。

How to understand, monitor, debug, and optimize agentic AI systems using OpenTelemetry and modern AI...

Three Decisions, Three Places Same agent. Same sandbox. Different folder. Different blast...