https://ai-tokens.cn/news/news-meta-muse-20260922.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Meta 个人 AI 智能体 Muse 上线两周登顶 App Store，单日市值飙升近 2000 亿美元 https://ai-tokens.cn/news/news-gpt-6-astra-20260903.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 OpenAI 发布 GPT-6 Astra：「欢迎来到 AGI 时代」，API 定价 $10/$50 为上代 2.5 倍 https://ai-tokens.cn/news/news-gemini-38-flash-20260903.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Google 三周内再推 Gemini 3.8 Flash：介绍价 $0.75/$3.75，Cyber 网络安全变体齐发 https://ai-tokens.cn/news/news-workbuddy-open-platform-20260903.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 WorkBuddy 开放平台解析：用 Skill、Connector、专家与知识库给 AI 同事「配工具」 https://ai-tokens.cn/news/news-wechatpay-ai-card-20260831.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 微信支付 AI 专属卡上新：一张卡直连 DeepSeek Harness 与 OpenClaw https://ai-tokens.cn/news/news-anthropic-mhs-20260828.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Anthropic 发布 MHS 模型硬件标准：让 AI 像插 USB-C 一样操控真实世界设备 https://ai-tokens.cn/news/news-f1e5a9e0a45e91bb.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 首款韬定律逻辑折叠芯片量产，麒麟9050让华为旗舰手机玩出什么新花样？ https://ai-tokens.cn/news/news-31640ef84cacf96f.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 “圳”启“芯”程 智创未来！2026湾芯展盛大启幕 https://ai-tokens.cn/news/news-4f4601cb94064617.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Figma announces new features to define how agents work on its platform https://ai-tokens.cn/news/news-72228f3d165edfb1.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Meet Grok Voice 3.0: SpaceXAI Voice Agent Model Ranks #1 on Artificial Analysis https://ai-tokens.cn/news/news-c9d380c0b10dcf21.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Artificiality's First Exam: An LLM-Referenced C-test for Supraintelligence https://ai-tokens.cn/news/news-45a31fcfe6d3c3fc.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 The Silent Failures in Agentic Data Science https://ai-tokens.cn/news/news-6d32cf3f2a6a67e3.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 ClawLoop: Self-Improving Agents via Harness-Native Training Loop https://ai-tokens.cn/news/news-c4eeefc4e76f2fa4.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Correction Training for Diffusion Models https://ai-tokens.cn/news/news-32b179101ee9f44a.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Long-Horizon-Terminal-Bench 2.0: Automating Diverse Task Construction and Measuring Sustained Agent Progress https://ai-tokens.cn/news/news-08d6ee7f7a96b8a9.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 VETNET: Assembling Benchmark Datasets for AI-Driven Veterinary Innovation https://ai-tokens.cn/news/news-fc3118d0cb024e58.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Unconstrained CTI-REALM with Open-Weights https://ai-tokens.cn/news/news-dc7bf030489751bc.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Visualizing the World Model in an Agent's Value Function https://ai-tokens.cn/news/news-728b011562c91b03.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Cross-Modal Latent Influence: Patch-Level Training Sample Attribution for Vision-Language Models https://ai-tokens.cn/news/news-020914d440160445.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Should SLMs be in the Driver's Seat? A Study of Local Models as Interpreters in Cloud-Written Agent Programs https://ai-tokens.cn/news/news-9e49e6d93707d1f6.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Harness^2: Recursive Agent Harnessing for an Open World https://ai-tokens.cn/news/news-f1022d584e2a85d3.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Echo: Error-Compensated 4-Bit Tensor-Parallel Communication for LLM Training and Inference https://ai-tokens.cn/news/news-7eef65c93bc57083.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Controllable Diverse Sampling through Pullback Directions https://ai-tokens.cn/news/news-79014dd89c495e69.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 ConicFlow: constrained Inference-time optimal control for flow matching models https://ai-tokens.cn/news/news-b84ea2ee2efdb648.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Which Metadata Matters for LLM Tool Selection in MCP? Evidence from Controlled Factor-Level Interventions https://ai-tokens.cn/news/news-6d86d47cf742e4db.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Following the Flow the Law Requires: Training Voice Agents to Transfer to Unseen Regulated Procedures in Debt Collection https://ai-tokens.cn/news/news-c690c3ae06d3d98a.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Evaluating Large Language Models for Multilingual Abstractive Summarization: A Comprehensive Study of English and Bengali https://ai-tokens.cn/news/news-ae48e1f386f597e7.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Can LLMs Access their World Knowledge for Event Prediction? https://ai-tokens.cn/news/news-56cbb1baf422b773.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Joint Subspace Sparse Attention for Efficient Long-Context Decoding https://ai-tokens.cn/news/news-984067f85ba96831.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Pre-Trained Language Models as Zero-Shot Tools for Social Science Research https://ai-tokens.cn/news/news-6fc7f72a81c622b7.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 From Copying to Composition: How Language Models Develop In-Context Learning https://ai-tokens.cn/news/news-e87076ca725dbb21.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 CAFE: Confining Feature Distillation to Language-Routed Adapters for Compact Multilingual Automatic Speech Recognition https://ai-tokens.cn/news/news-95ca3dc5933b9adf.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Fantastic Metrics and How to Find Them: Meta-Evaluating Metrics for Multi-lingual Style-Personalized Generation https://ai-tokens.cn/news/news-93bf7dcf4335a1e6.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 HERO: Harnessing Evidence for Test-time Refinement and Prompt Optimization in Program Generation https://ai-tokens.cn/news/news-2c2dfc33228ae5fc.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 How to Evaluate When You Don't Know How to Evaluate? Choosing Reliable Setups for Base Language Model Benchmarking https://ai-tokens.cn/news/news-c35ab325892985e9.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 CLASHarness: Benchmarking Multi-Agent Conflict in Computer Use via a Controllable and Verifiable Generation Pipeline https://ai-tokens.cn/news/news-4b851de069c29b77.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Why Do Agent Teams Fail Across Boundaries? A Failure Taxonomy for Agent-to-Agent Systems with Private States https://ai-tokens.cn/news/news-ec2fe4aeb931678e.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 XferAgent: Cross-Agent Experience Transfer for Self-Improving LLM Agents https://ai-tokens.cn/news/news-304f78285564b498.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 One Agent per Zone: Decentralised Active Inference under Partial Observability for Multi-Zone HVAC Control on Edge Hardware https://ai-tokens.cn/news/news-a74a082b6b7bb007.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 When Extremists Enter LLM Societies: Investigating Extreme Behavior in Multi-Agent Systems https://ai-tokens.cn/news/news-ed70a6c087a3ec66.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 From Training to Escalation: Forensic Analysis and Local Recreation of RL/RLHF-Induced Privilege Escalation in AI Agents (CVE-2026-65616) https://ai-tokens.cn/news/news-ac4956945d4b2452.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 CIPHEUR: LLM-Guided Structural Search Control for Conflict-Graph Optimization https://ai-tokens.cn/news/news-c643418cbbaee3ee.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 CyberArena: Benchmarking AI Agents in Live Symmetric Attack-and-Defense Combat https://ai-tokens.cn/news/news-4aa0285e48f2d54a.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Smart Alone, Blind Together: Why Accuracy Fails to CertifyTheory of Mind in Multi-Agent Teams https://ai-tokens.cn/news/news-8b93229840889bda.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Narrowing the Collaboration Gap, Probably https://ai-tokens.cn/news/news-960b155c4ff100a8.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Truthful and Expected-Payment-Optimal Model Routing in Graph-Based Agentic Workflows https://ai-tokens.cn/news/news-a3e0b7846ea18751.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 When Is Smaller Greener? Right-Sizing Language Models for Agentic AI https://ai-tokens.cn/news/news-29f47425d7a11074.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 AdCoEvo: A Co-Evolving Multi-Agent Advertising Delivery System https://ai-tokens.cn/news/news-7eb214d807f45e74.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Lotus: Learning to Understand Time-series by Learning to Forecast https://ai-tokens.cn/news/news-0d0e88ba1d1de519.html AI导航 zh-CN 2026-10-15T14:58:21+08:00 Inverse Reasoning Models for Efficient Post-Training and Verification from Answer-Only Data