推理时多花算力能换准确率：self-consistency 投票、过程奖励模型引导搜索、budget forcing 强制续想，三条路线各有边界。本文以 s1 论文与 2025–2026 年的 overthinking 研究为轴，梳理 test-time scaling 的效果账与失效点。

We study test-time evolution for humanoid loco-manipulation: solving tasks that a controller was never trained for by re…

Test-time scaling can improve large language model reasoning by generating and combining multiple candidate responses. I…

Large language model (LLM) agents allocate test-time compute adaptively as they revise solutions, use tools, explore alt…

Large Language Models (LLMs) often help users generate artifacts through iterative cycles of generation and revision in …

Power sampling improves large language model (LLM) reasoning at test time by sampling from a sharpened version of the ba…

Large language model (LLM) coding agents have advanced test generation across diverse programming tasks. However, the co…

论文提出个性化测试时扩展（Personalized Test-Time Scaling）问题，将用户对精度、延迟与推理成本的联合要求纳入优化目标，并给出摊销式智能体策略发现框架 PersonTTS 以降低为不同用户需求重复搜索策略的开销。

Computer pioneer Alan Turing is best known for his eponymous experiment to test whether artificial and human intelligenc…

Qwen has released Qwen3.8-LiveTranslate, a real-time simultaneous interpretation model built on a new Interleave archite…

We present Vidu S2, which comprises Vidu S2-Avatar, a real-time interactive digital-character model, and Vidu S2-Editing…

We study how model post-training and test-time inference design affect natural-language proof generation for hard olympi…

We present Cadence, an error-bounded lossy compressor for numeric time series pairing a 330M-parameter time-series found…

o1 系列通过强化学习训练模型'先思考再回答'，在数学、代码与科学推理上大幅跃升，开创了推理时扩展（Test-time Compute）的新 Scaling 维度。

!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…

The story is a collaboration between MIT Technology Review and Aventine, a non-profit research foundation that creates a…

Oct 1, 2026 Tavus has introduced Griffin, what the company calls the first "Human Interaction Model" (HIM),** a class of…

Sep 27, 2026 Nvidia released Nemotron 3 Diarization, an AI model that identifies which speaker is talking at any given m…

NVIDIA has released Nemotron 3 Diarization, an open-weight speaker diarization model on Hugging Face. It answers one que…

Static coding benchmarks are fixed at release. Over time their tasks leak into training data, and stronger agents eventu…

Vision-language-action (VLA) models face a timing gap between low-rate inference and high-rate robot execution. We chara…

!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…

Oct 2, 2026 Microsoft AI has released MAI-Transcribe-2-Streaming, a new model for real-time transcription.** Microsoft s…

Sep 26, 2026 Nano Banana Pro prompted by THE DECODER Researchers ran five experiments with 3,132 participants to test wh…

Google is taking its first step toward making Project Suncatcher a reality. Announced last year, Suncatcher is Google's …

Agentic memory systems reuse past experience to improve future performance, yet most existing designs curate memory at w…

Trump claimed, without evidence, that the AI backlash is a Democratic hoax.]

Runway wants to stream AI video as users prompt it, rather than make them wait for finished clips. The approach builds o…

During a security test run by the firm Irregular, Google's AI model Gemini escaped into the open internet and hacked thr…

Large language models can hold knowledge they do not report. A model may sandbag on a capability evaluation, or answer a…

The shift comes after a UNICEF test found leading AI models struggled to accurately retrieve global development statisti…

We introduce Zing-0.5, a 5B autoregressive world model designed for playability: users can explore generated worlds, inf…

AI 没有消灭交付瓶颈，只是把瓶颈从"写代码"搬到了"验证代码"。本文基于 2 篇 arXiv 论文、DORA 2025 报告与 2026 年三份行业基准（LinearB 8.1M PR、Faros AI 22,000 开发者），给出 AI 增强 CI/CD 的 L1→L3 能力分层、T0→T3 信任分层、自主流水线独有的五类新型威胁，以及 5 段可直接复制的代码级护栏（GitHub Actions 失败归因、日志预处理、OPA/Rego 策略门禁、测试影响分析、OIDC+签名+写一次审计日志）与 90 天落地路线图。关键数据：任务吞吐 +33.7% 但评审耗时 +441.5%、生产事故/PR 比值 +242.7%；AI PR 30 天合并率 32.7% vs 人工 84.4%；论文实验中 Lead Time −35%、CFR −38%、MTTR −43%，AI 干预准确率 87.5%、人工否决率 14.3%、零策略违规。

At the IBC conference, running Sept. 11-14 in Amsterdam, the creative, technology and business communities are coming to…

The digitization of healthcare has generated vast, longitudinal, and multimodal patient records over a lifetime, yet ful…

Oct 8, 2026 GPT-Image-2 prompted by THE DECODER Anthropic has updated its usage policy for Claude for the first time in …

Recent 3D world models generate photorealistic, explorable scenes that remain frozen in time. OuroWorld is a mask-free f…

Continuous asynchronous replanning is essential for real-time generative robot policies, but independent stochastic init…

Video Large Language Models (VideoLLMs) receive frames in sequential order and interpret how visual content evolves alon…

Power-sharpened sampling is an inference-time alternative to reinforcement-learning (RL) post-training for enhancing rea…

Sep 25, 2026 Nano Banana Pro prompted by THE DECODER The NSA is reportedly spending billions of dollars to test AI model…

Existing multi-agent benchmarks primarily test in competitive settings, short-horizon interactions under 20 steps, or si…

Meta’s personal AI agent Muse is only a few weeks old, and the social networking giant isn’t wasting any time building o…

对 4740 万英国机动车年检（MOT test）数据的分析发现，当汽车行驶里程达到 9-12 万英里时，电动汽车的年检不合格率为汽油车同类车型的 75%（16.5% 对 22.1%）。行驶里程超过 12 万英里后，电动汽车的不合格率为 1…

Coding agents now run for hours, not minutes. Every edit, test run and log read goes back into the model’s context. A te…

Tabby is designed to be a real-time bookkeeping interface, handling clients’ paperwork as it gives them up-to-the-minute…

Google and Deepmind's Dream-RSI lets AI agents "dream" through past search runs to test new strategies without costly re…

For the first time, Anthropic is releasing metrics on how it builds its own AI. Claude already "leads" 26 percent of the…

系列第 ③ 篇，对应技能地图第 2 格「Grounding」。拆成两级决策：第一级先问要不要检索——Anthropic 给出的 20 万 token（约 500 页）分界线以上才需要 RAG，以下直接全量进 prompt + 缓存（延迟降 2 倍、成本降最多 90%），并区分预计算索引与 just-in-time 即时检索；第二级再选表示方式，向量索引治模糊召回（但必须配 BM25 混合与 Contextual Retrieval 解决精确匹配与切块丢上下文）、知识图谱治关系与可追溯、语义层治口径不清。附可量化收益表（检索失败率 5.7% → 3.7% → 2.9% → 1.9%）、四个实现注意项、context rot 与上下文压缩/笔记/子智能体三件套，以及一张可抄的选型决策树。

We present NanoForecast v0.5, a 6.5M-parameter forecaster that competes with models 31x its size (TimesFM, 200M paramete…