OpenAI2026-09-25新兴中
Proaction boosts sales 60% and saves 75+ hours with Codex
TL;DRProaction boosts sales 60% and saves 75+ hours with Codex
AI Learning Lab / 概念账本
账本在生长，半径不变
定时抓取 OpenAI / Google DeepMind / Hugging Face / arXiv 等一手来源，并策展 Meta Muse / Kimi / Creao AI / Genspark 等技术文章。每张卡片是一条可对读的展陈，不是信息流。
自上次访问新增 28 个版本记号（v28）。
订阅概念方向选择关心的方向；新概念以站内横幅提醒
01按控制半径浏览
与首页知识网络、基础馆章节共用同一组节点。点选即筛选账本。
02检索与筛选
03账本
共 400 条
Anthropic2026-09-24新兴中
Claude discovers a novel enzyme system
TL;DRIn early results from our new life sciences research lab, Claude agents found an enzyme sy…
Google DeepMind2026-09-24新兴中
Introducing Gemini 3.8 Live with Live Avatar
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Hugging Face2026-09-24新兴中
Accelerating vision-language models with LFM2.5-VL-DSpark
TL;DRHow does speculative decoding work for VLMs The vision drafter uses the same architecture…
arXiv AI2026-09-24提出中
LLM Agents Can Easily Tamper With Their Own Traces
TL;DRAsynchronous monitoring, incident investigations, and compliance audits primarily rely on…
arXiv AI2026-09-24提出中
AD-WM: Action-Discriminative World Models for Counterfactual Model Predictive Control
TL;DRLatent world models are typically trained to predict factual transitions, whereas model pr…
arXiv AI2026-09-24提出中
Agentic Detection of Online Conspiracies
TL;DRConspiratorial discourse on social media is not always expressed through explicit claims o…
arXiv AI2026-09-24提出中
RAPID: Robot Agentic Programming from Demonstrations
TL;DRCoding agents have demonstrated enormous success in solving complex programming problems.…
arXiv AI2026-09-24提出中
Rolling-WAM: World Action Models with Rolling Imagination
TL;DRWorld Action Models (WAMs) couple action generation with future visual prediction for robo…
arXiv AI2026-09-24提出中
JevOut: Natural Context Can Flip Decision Models
TL;DRDedicated decision models such as Jev map unstructured language to probability distributio…
arXiv AI2026-09-24提出中
SemMSA: Latent Semantic-Aided Robust Multimodal Sentiment Analysis with Incomplete Data
TL;DRRecent research on Multimodal Sentiment Analysis (MSA) has focused on learning from langua…
arXiv AI2026-09-24提出中
Coding Agents for Generalized Task and Motion Planning Problems
TL;DRTask and motion planning (TAMP) problems remain difficult even with full observability and…
arXiv AI2026-09-24提出中
To Trust or Not to Trust: Retrieval-Augmented Fact Checking in Speech
TL;DROnline misinformation increasingly appears in spoken formats such as news clips, podcasts,…
arXiv AI2026-09-24提出中
PoEM: Predicting RL Outcomes from Existing Policies
TL;DRFoundation models are post-trained with reinforcement learning (RL) to maximize specific r…
arXiv AI2026-09-24提出中
TrackEverything: Long Horizon Dense Tracking via De-Duplicating 3D Scene Representations
TL;DRExisting point tracking models face a fundamental tradeoff: they can either track a sparse…
arXiv AI2026-09-24提出中
Requirement-Bound Verified Commissioning: A Frozen Four-Billion-Parameter Local Model as a Candidate Generator under an External Acceptance Layer with Verification and Release Authority
TL;DRAn acceptance protocol is developed for sensor-coordinate and polarity binding in mechatro…
arXiv AI2026-09-24提出中
Minimally Invasive Steering of Language Models
TL;DRPre-logit steering adapts a frozen language model to a test-time reward by adding vectors…
arXiv AI2026-09-24提出中
Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure
TL;DRA central concern in AI safety is that agents may treat oversight as an obstacle when it c…
arXiv AI2026-09-24提出中
Underwater C3-JEPA: An Object-Centric Cross-View World Model for ROV Salvage
TL;DRWe present Underwater C$^{3}$-JEPA (cross-view, control-conditioned, context-extended), an…
arXiv AI2026-09-24提出中
A Living Benchmark for Information Retrieval from Electronic Health Records
TL;DRLarge language model (LLM)-based clinical assistants are increasingly being integrated int…
arXiv AI2026-09-24提出中
ExplorationBench: Measuring AI Systems' Exploration in Verifiable Alien Worlds
TL;DRScientific discovery begins where known problems end. There, AI systems must engage in exp…
arXiv AI2026-09-24提出中
SAGE: Mitigating Long-Horizon Reasoning Biases via Topological Guidance
TL;DRLong-horizon reasoning remains a central challenge for large language models (LLMs) under…
arXiv AI2026-09-24提出中
Jev-Mobile: Jev as an Executor for Mobile GUI Agents
TL;DRVision-language models (VLMs) have become a common foundation for autonomous mobile GUI ag…
arXiv AI2026-09-24提出中
ARGUS: Role-Aware Event Knowledge Graphs for U.S. Employment-Discrimination Complaints
TL;DRU.S. employment-discrimination complaints describe complex event sequences that are not ex…
Meta2026-09-24新兴中
Bringing Private Processing to Meta AI Glasses
TL;DRWe believe glasses are the best form factor for having AI help throughout your day. They c…
Google Gemini2026-09-24新兴中
Introducing Gemini 3.8 Live with Live Avatar
TL;DRIntroducing Gemini 3.8 Live with Live Avatar, which brings near real-time visual presence…
OpenAI2026-09-23新兴中
OpenAI extends cyber access to Ukraine for civilian defense
TL;DROpenAI extends cyber access to Ukraine for civilian defense
OpenAI2026-09-23新兴中
Grab and OpenAI bring practical AI skills to Southeast Asia
TL;DRGrab and OpenAI bring practical AI skills to Southeast Asia
OpenAI2026-09-23新兴中
Two years of OpenAI Academy
TL;DRTwo years of OpenAI Academy
OpenAI2026-09-23新兴中
Sam Altman’s remarks at the United Nations Security Council
TL;DRSam Altman’s remarks at the United Nations Security Council
OpenAI2026-09-23新兴中
Harvey turns legal context into stronger drafts with GPT-6 Astra
TL;DRHarvey turns legal context into stronger drafts with GPT-6 Astra
OpenAI2026-09-23新兴中
How invideo improves color grading 3x with GPT‑6 Astra
TL;DRHow invideo improves color grading 3x with GPT‑6 Astra
OpenAI2026-09-23新兴中
Ringg’s AI agents resolve up to 65% of customer calls with OpenAI
TL;DRRingg’s AI agents resolve up to 65% of customer calls with OpenAI
OpenAI2026-09-23新兴中
Introducing MentalHealthBench
TL;DRIntroducing MentalHealthBench
OpenAI2026-09-23新兴中
ChatGPT Ads expands to Southeast Asia and Taiwan
TL;DRChatGPT Ads expands to Southeast Asia and Taiwan
OpenAI2026-09-23新兴中
Airbnb widens access to GPT-6 Astra and OpenAI frontier models
TL;DRAirbnb widens access to GPT-6 Astra and OpenAI frontier models
Google DeepMind2026-09-23新兴中
Advancing Private AI Compute with secure, server-side memory
TL;DRIntroducing private, server-side memory to Private AI Compute for personal AI.
Google DeepMind2026-09-23新兴中
Gemini 3.8 text-to-speech says hello
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Google AI Blog2026-09-23新兴中
Google Beam expands with new regions, partners, and customers
TL;DRWe’re expanding Google Beam to five new countries, and partnering with Industrious for an…
Hugging Face2026-09-23新兴中
**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**
TL;DRConsider a transcript from a meeting, customer call, or podcast in which every sentence is…
Hugging Face2026-09-23新兴中
How to Use NVIDIA Warp and MjWarp to Accelerate Robotics Simulation and Learning Workflows
TL;DRMuJoCo Warp (MJWarp), built on NVIDIA Warp, takes compatible MuJoCo models into that GPU-s…
Google Gemini2026-09-23新兴中
A new wave of Connected Apps is rolling out to Gemini.
TL;DRGemini is adding new Connected Apps, including Adobe, Airtable, Linear, Peloton and more t…
Google Gemini2026-09-23新兴中
Gemini 3.8 text-to-speech says hello
TL;DRGemini 3.8 Flash-Lite TTS and Gemini 3.8 Flash TTS are our most expressive audio models ye…
arXiv AI2026-09-23提出中
Contrastive Learning for Authorship Verification
TL;DROur results show that contrastive learning outperforms a classification-based approach to…
arXiv AI2026-09-23提出中
StudentBench: AI and human tutoring yield equivalent GRE learning gains
TL;DRArtificial intelligence offers an unprecedented opportunity to augment human capabilities,…
arXiv AI2026-09-23提出中
Where Should I Join? Robot Group Joining via Language-Guided Goal Prediction
TL;DRSocial navigation typically assumes a specified goal and focuses on reaching it while resp…
arXiv AI2026-09-23提出中
Can LLMs Reason About Runtime Behavior? A Repository-Level Dynamic Benchmark
TL;DRLarge language models (LLMs) are increasingly used in coding tasks, but their ability to r…
arXiv AI2026-09-23提出中
Order-Invariant Answers, Order-Sensitive Representations in Mathematical Reasoning
TL;DRReordering a set of mathematical rules without changing its meaning should preserve the co…
arXiv AI2026-09-23提出中
Cross-Scale Transfer Learning for Depression Severity Prediction: From PHQ-8 to HAMD-17 Across Languages and Clinical Paradigms
TL;DRThis work addresses continuous depression-severity score prediction from clinical intervie…
arXiv AI2026-09-23提出中
Agent-Editing World Model: Rethinking World Modeling for LLM Agents
TL;DRRecent advances in large language models (LLMs) have enabled agents to tackle long-horizon…
arXiv AI2026-09-23提出中
Frozen Flows Forget: Diagnosing and Restoring Lost Motion in a Latent-flow World Model
TL;DRLatent world models that integrate a flow in a frozen self supervised latent space train s…
arXiv AI2026-09-23提出中
Learning Holographic Reduced Representations with Clifford Variational Autoencoders
TL;DRVector Symbolic Algebras project data structures into a hyperdimensional vector space thro…
arXiv AI2026-09-23提出中
Fine-Tuning LLMs for Translation: General Forgetting Mitigation Does Not Preserve MT-Specific Instruction Following
TL;DRFine-tuning large language models on parallel data improves translation quality but can ca…
arXiv AI2026-09-23提出中
When and Where to Trust the Teacher: Unifying On-Policy Distillation and GRPO through Entropy-Calibrated Credit Assignment
TL;DRReinforcement learning with verifiable rewards (RLVR) supervises mathematical reasoning th…
arXiv AI2026-09-23提出中
Shopping by algorithm: How agentic AI deploys human heuristics as a surrogate consumer
TL;DRConsumers increasingly delegate purchasing decisions to Large Language Models (LLMs) actin…
arXiv AI2026-09-23提出中
AnchorReasoning: A Visual Grounding and Causal Reasoning Dataset in Long-Tail Autonomous Driving Scenarios
TL;DRVision-language models (VLMs) offer a promising approach to long-tail autonomous driving,…
arXiv AI2026-09-23提出中
MicroQonv: Reshaping Convolution Tensors for Efficient Microscaling in Training and Inference
TL;DRMicroscaling quantization techniques are increasingly used to represent neural network par…
arXiv AI2026-09-23提出中
Digital diglossia: Arabic between X and Facebook
TL;DRThis study highlights the distribution of Standard Arabic (SA; H(igh) variety) and Colloqu…
arXiv AI2026-09-23提出中
Mizar: A 159M-Parameter Audio-Language Model for Audio Understanding
TL;DRAudio-language models (ALMs) integrate acoustic perception with the knowledge encoded in l…
arXiv AI2026-09-23提出中
An Open Pipeline and Dashboard for Systemic-Risk Evidence under the EU AI Act's Code of Practice
TL;DRClaims about AI safety reach audiences well beyond the AI community, yet many rely on opaq…
arXiv AI2026-09-23提出中
Learning the Cost of Reliable Inference
TL;DRBenchmarking and routing platforms increasingly act as intermediaries connecting large lan…
arXiv AI2026-09-23提出中
Computation Over Geometry: Meaning Identity Is Computed, Not Shipped in the Embeddings
TL;DRMeaning identity (whether two sentences say the same thing after wording changes) is treat…
arXiv AI2026-09-23提出中
Shutdown Sabotage Propensities in Multi-Agent Systems
TL;DRThe final safeguard against rogue AI behavior is the human ability to shut systems down. I…
OpenAI2026-09-22新兴中
Better prompt caching for GPT-6
TL;DRBetter prompt caching for GPT-6
OpenAI2026-09-22新兴中
Introducing GPT-6 Sol and Luna
TL;DRIntroducing GPT-6 Sol and Luna
OpenAI2026-09-22新兴中
Parallel cut research time and cost in half with GPT‑6 Astra
TL;DRParallel cut research time and cost in half with GPT‑6 Astra
OpenAI2026-09-22新兴中
Priorities and principles for effective third party assessments
TL;DRPriorities and principles for effective third party assessments
Hugging Face2026-09-22新兴中
Transformers now runs llama.cpp quants
TL;DRRunning AI models on your laptop has become much easier, and llama.cpp has been a big part…
Hugging Face2026-09-22新兴中
Jun Kim, oMLX creator and maintainer, joins Hugging Face to support the MLX community
TL;DRMLX is Apple's framework for local AI, especially optimized for Apple Silicon. We are big…
Hugging Face2026-09-22新兴中
How UK AISI and EvalEval Are Making Benchmark Results Reproducible
TL;DRAISI and EvalEval have previously collaborated on research that began at a joint workshop…
arXiv AI2026-09-22提出中
Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs
TL;DRDiffusion Large Language Models (dLLMs) have recently emerged as a promising alternative t…
arXiv AI2026-09-22提出中
Agensh: Scaling Organizational Intelligence to 1,024 Agents
TL;DRA multi-agent system can reduce latency on complex tasks by executing work concurrently. S…
arXiv AI2026-09-22提出中
SpeakerMem-R1: Speaker-Centered Dual-Track Memory for Multi-Party Dialogue
TL;DRLong-term conversational memory in multi-party settings requires more than retrieving rele…
arXiv AI2026-09-22提出中
CliffCompaction: Cost-Efficient Compaction for Long-Horizon Coding Agents
TL;DRAgents often work on complex problems that require millions of tokens of context, which ne…
arXiv AI2026-09-22提出中
SWE-Serve: Benchmarking Agentic Engineering For Production Inference Serving
TL;DRWe introduce SWE-Serve, a benchmark for evaluating agents on production inference engineer…
arXiv AI2026-09-22提出中
A2M: Trace-Optimized Agent Hijacking in the MCP Ecosystem
TL;DRAgents using the Model Context Protocol (MCP) rely on semantic matching to select tools fr…
arXiv AI2026-09-22提出中
Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents
TL;DRLarge language model (LLM) agents often handle streams of related tasks, yet standard harn…
arXiv AI2026-09-22提出中
Type-Safe Is Not Error-Free: A Constrained Decision Head Follows the Option Name, Not the Rubric Bound to It
TL;DRTyped decision models are built for settings where model outputs are consumed directly by…
arXiv AI2026-09-22提出中
FleXray: Universal Clinical X-ray Segmentation
TL;DRX-ray is medicine's most widely used imaging modality, yet remains among its least quantit…
arXiv AI2026-09-22提出中
Metrics Failure in LLM-Based Code Vulnerability Repair: An Empirical Study and a Change-Aware Screen
TL;DRLarge language models (LLMs) are increasingly applied to the automated repair of C/C++ sec…
arXiv AI2026-09-22提出中
Does AI Save Time on Product Design? A Randomized Controlled Experiment of AI Prompt-to-Design Workflows
TL;DRAI tools for digital product design now offer prompt-to-design capabilities, allowing desi…
arXiv AI2026-09-22提出中
The Sirens' Song: When Proximal Background Context Overshadows Distant Evidence
TL;DRLong-context LLMs focus on retrieving distant evidence from extensive context, yet existin…
arXiv AI2026-09-22提出中
TraceVIC: Causal Reasoning over Code Evolution for Identifying Vulnerability-Inducing Commits
TL;DRSoftware vulnerabilities are often discovered long after they are introduced, making it di…
arXiv AI2026-09-22提出中
Train Where the Quantized Model Goes: On-Policy Distillation for Low-Bit Reasoning
TL;DRQuantization-aware distillation (QAD) restores much of the short-form question-answering p…
arXiv AI2026-09-22提出中
Beyond Repeated Sampling: Learning Search Policies for LLM Reasoning
TL;DRLarge language models increasingly tackle hard reasoning problems by spending more test-ti…
arXiv AI2026-09-22提出中
Measuring the Serving Stack Instead of the Model: Hidden Confounds in Local Tool-Use Evaluation
TL;DRA coding agent must emit a valid tool call--a parseable invocation of a tool in the provid…
arXiv AI2026-09-22提出中
Detecting GPT-Assisted Writing Using Interpretable Stylometric Features
TL;DRDistinguishing GPT-assisted from independently authored student writing has become a criti…
arXiv AI2026-09-22提出中
From Alignment to Access Control: A Framework for GenAI Policy Enforcement
TL;DRGenerative AI (GenAI) applications have flourished enabling users to chat with large langu…
arXiv AI2026-09-22提出中
A Spectral Theory of Grokking: Weight Decay induces Feature Learning
TL;DRIn grokking an early fit to the training data separates from a much later improvement in g…
arXiv AI2026-09-22提出中
Discovery-Driven Integration of Disjoint Tables via Text
TL;DRIntegrating heterogeneous datasets within data lakes is a critical challenge, particularly…
Hugging Face2026-09-21新兴中
Pruning LLMs Like a Physicist: Block Removal as an Ising Optimization Problem
TL;DROur latest paper, LLM Compression by Block Removal with Constrained Binary Optimization, t…
Hugging Face2026-09-21新兴中
tokenizers v1: encode, decode and scaling, measured
TL;DRAs models become faster and workloads scale, that balance begins to shift. Training on mas…
Meta2026-09-21新兴中
Open-Sourcing Rebalancer: A Generic, High-Performance Library for Solving Assignment Problems
TL;DRWe’re open-sourcing Rebalancer, the assignment-problem solver that has been used to solve…
Meta2026-09-21新兴中
Inside Petal: Building the World’s First Petabit-Class Transoceanic Subsea Cable
TL;DRPetal, the next step in Meta’s subsea innovation, will be the first subsea cable to delive…
OpenAI2026-09-21新兴中
How V7 gives AI agents institutional memory
TL;DRHow V7 gives AI agents institutional memory
OpenAI2026-09-21新兴中
Advisory Group on Mathematics and Artificial Intelligence
TL;DRAdvisory Group on Mathematics and Artificial Intelligence
OpenAI2026-09-21新兴中
Higgsfield AI ships new video features in a day with GPT-6 Astra
TL;DRHiggsfield AI ships new video features in a day with GPT-6 Astra
OpenAI2026-09-21新兴中
Building standards for the next phase of AI
TL;DRBuilding standards for the next phase of AI
OpenAI2026-09-21新兴中
Expanding OpenAI Academy with new learning paths
TL;DRExpanding OpenAI Academy with new learning paths
arXiv AI2026-09-21提出中
GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay
TL;DRModern video games provide a measurable testbed for AI models, combining abilities of visu…
arXiv AI2026-09-21提出中
Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use
TL;DRMulti-turn tool-use failures can hinge on a single model call, yet reward variation alone…
arXiv AI2026-09-21提出中
WorldCrafter: Consistent Video World Model with Implicit 3D-aware Memory
TL;DRVideo world models enable interactive exploration of dynamic environments, yet struggle to…
arXiv AI2026-09-21提出中
onPanda: Efficient Annotation of On-Policy Alignment Data for LLMs and Agents via Token-Level Correction
TL;DRWe present onPanda, an interactive tool for efficiently annotating LLM alignment data and…
arXiv AI2026-09-21提出中
DexTacWAM: A Visuo-Tactile World-Action Model for Dexterous Manipulation
TL;DRDexterous manipulation depends on contact dynamics that are often only partially observabl…
arXiv AI2026-09-21提出中
Harness-Zero: Harness Distillation via Agent-as-Harness
TL;DRAgent harnesses, the external systems that mediate model-environment interaction, can subs…
arXiv AI2026-09-21提出中
RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
TL;DRAn LLM agent's capability is largely magnified by its harness, namely the prompts, control…
arXiv AI2026-09-21提出中
DolphinBench: Mapping the Pareto Frontier of Agent Memory
TL;DRAgents today often take real-world actions that depend on long-term memory and context rec…
arXiv AI2026-09-21提出中
Rare Event Estimation via Iterative Unalignment
TL;DRAs agents are deployed with increased autonomy, even extremely rare events along their sto…
arXiv AI2026-09-21提出中
Emergent Collusion in Long-Horizon LLM Agent Interaction
TL;DRLLM agents are increasingly deployed in collaborative settings, yet long-term interaction…
arXiv AI2026-09-21提出中
Jev for Scientific Decisions: Evaluating Semantic Choices and Their Consequences
TL;DRScientific workflows often require choosing among known relations before a deterministic c…
arXiv AI2026-09-21提出中
Generative Tutorial: Towards Live Contextualized Visual Instructions for Physical Tasks
TL;DRVisual instructions for physical tasks are typically authored in one context and followed…
arXiv AI2026-09-21提出中
Exactness at Inference: A Representational Criterion for Out-of-Distribution Generalization
TL;DRA model generalizes outside its training distribution only when it computes a representati…
arXiv AI2026-09-21提出中
Linguistic Features for Interpretable Textual Entailment
TL;DRDespite the success of neural models in natural language processing, their black-box natur…
arXiv AI2026-09-21提出中
Et Tu, Brute? Economic Misalignment in Personal AI Agents
TL;DRPersonal AI agents make recommendations and take actions on people's behalf in high-stakes…
arXiv AI2026-09-21提出中
BackTrend: Evaluating Scientific Weak-Signal Prediction via Backward Reconstruction
TL;DRScientific weak signals are early, low-visibility research directions that later become ce…
arXiv AI2026-09-21提出中
SocioVerse2: A Longitudinal Dynamic Social Simulation Framework under a Human-AI Co-evolutionary Paradigm
TL;DRSocial simulation offers the social sciences an experimental instrument that the real worl…
arXiv AI2026-09-21提出中
Visuomotor Robotic Pruning in Planar Orchards Using Hybrid Reinforcement Learning
TL;DRDormant tree pruning is labor-intensive yet essential for maintaining modern high-producti…
arXiv AI2026-09-21提出中
ToneCL: Contrastive Learning for Few-Shot Syllable-Level Tone Classification
TL;DRTone languages constitute over 50-70% of the world's languages, but the vast majority are…
arXiv AI2026-09-21提出中
Human-LLM Deliberation as Interactive Proof: Conditions for Verifiability Without Transparency
TL;DRWhen an LLM supplies an argument that a user could not readily construct, how can the user…
Anthropic2026-09-18新兴中
Partnering with Accenture on embedded evaluation
TL;DRWe’re partnering with Accenture on independent evaluation of frontier AI—part of our recen…
Google AI Blog2026-09-18新兴中
New experts join Google’s AI & Economy team
TL;DRWe are expanding our AI & Economy team with world-class academic advisors, fellows, and co…
Google AI Blog2026-09-18新兴中
Co-creating the future of fashion with Google
TL;DRGoogle worked side-by-side with designers Jane Wade and Sergio Hudson to custom-design Goo…
OpenAI2026-09-18新兴中
Introducing the Australian Youth Safety Blueprint
TL;DRIntroducing the Australian Youth Safety Blueprint
arXiv AI2026-09-18提出中
Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design
TL;DRProfessional graphic design is a long-horizon agentic task in which structured, editable a…
arXiv AI2026-09-18提出中
Cross-sector generalization of accident-process role classification in occupational accident narratives
TL;DROccupational accident narratives contain valuable information about work situations, unfav…
arXiv AI2026-09-18提出中
CodeMidas: Scaling Agentic Coding RL Environments from Code Itself
TL;DRTraining capable coding agents via reinforcement learning (RL) requires diverse tasks with…
arXiv AI2026-09-18提出中
Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw
TL;DRUsers increasingly delegate work to autonomous AI agents, yet evaluations typically measur…
arXiv AI2026-09-18提出中
Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention
TL;DRMulti-hop retrieval failures are not uniformly distributed across queries: they cluster in…
arXiv AI2026-09-18提出中
An Interpretable Memory Decision Controller for LLM Agents Based on Three-Signal Complementarity: Decoupling Confidence and Consistency
TL;DRMemory systems for large language models have focused predominantly on efficient retrieval…
arXiv AI2026-09-18提出中
Gricea: An Open Science Platform for Conversational AI Research
TL;DRWe need studies on conversational AI (CAI) at scale to understand human behavior and shape…
arXiv AI2026-09-18提出中
QuranicMMLU: A Cognitively-Aware Benchmark for Evaluating Generative AI Solutions on Quranic Linguistic Knowledge
TL;DRWe introduce QuranicMMLU, a benchmark for evaluating generative AI on Quranic Arabic acros…
arXiv AI2026-09-18提出中
DiaVLo: Diagnosing Behaviours of Vision-Language Models
TL;DRVision-language models (VLMs) rely on storing and transferring appropriate information acr…
arXiv AI2026-09-18提出中
Abstention and Noise Filtering: Two Missing Primitives of Softmax Attention
TL;DRGating the value pathway of attention reportedly improves language model pretraining, and…
arXiv AI2026-09-18提出中
RecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use Agents
TL;DRComputer-use agents (CUAs) have advanced along two separate lines: graphical interaction a…
arXiv AI2026-09-18提出中
Bayesian Belief Layer for Controllable Opinion Dynamics in LLM Agents
TL;DRLLM agents in social simulation revise their opinions implicitly, in context: how open an…
arXiv AI2026-09-18提出中
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal
TL;DRLarge language models can hold knowledge they do not report. A model may sandbag on a capa…
arXiv AI2026-09-18提出中
Moral Entropy: Auditing Bias and Uncertainty in Moral Judgment
TL;DRMost work in computational ethics treats annotator disagreement on moral content as noise…
arXiv AI2026-09-18提出中
NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities
TL;DRWe introduce NemotronLabs VoiceChat, an open full-duplex speech-to-speech model with nativ…
arXiv AI2026-09-18提出中
Learning Cardiac Features: ECG Biometrics Across Time and~Exercise
TL;DRElectrocardiograms (ECGs) carry subject-specific patterns enabling reliable individual dis…
arXiv AI2026-09-18提出中
When Should a Failing Robot Ask? Initiating Corrective Human-Robot Dialogue from Audited Sensor Evidence
TL;DRA robot that fails at a task faces the first decision in corrective dialogue: act on its o…
arXiv AI2026-09-18提出中
AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory
TL;DRLong-term memory is essential for large language model (LLM) agents to maintain consistenc…
arXiv AI2026-09-18提出中
What Should We Ask Next? Retrieval-Aware Question Learning under Partial Evidence
TL;DRInteractive retrieval under partial evidence is a sequential information-acquisition probl…
arXiv AI2026-09-18提出中
Detecting Pretraining Data in Large Language Models from a Free-Energy Perspective
TL;DRDetecting pretraining data in large language models is challenging because high likelihood…
Anthropic2026-09-17新兴中
Introducing the Life Sciences Verification Program
TL;DRAnthropic is an AI safety and research company that's working to build reliable, interpret…
Google AI Blog2026-09-17新兴中
Making global data easier to explore
TL;DRGoogle and the UN system have launched the UN System Data Commons, a new open platform mak…
OpenAI2026-09-17新兴中
How Cooley is accelerating IPO work with ChatGPT
TL;DRHow Cooley is accelerating IPO work with ChatGPT
OpenAI2026-09-17新兴中
Introducing Astra for Law
TL;DRIntroducing Astra for Law
arXiv AI2026-09-17提出中
Coding Agents with an Obstacle-Aware Harness for Safe Robot Manipulation
TL;DRCoding agents have emerged as a promising paradigm for robot manipulation: a language mode…
arXiv AI2026-09-17提出中
Embedding Models Measure in Peculiar Ways
TL;DREmbedding spaces define notions of semantic similarity and distance. We study whether thos…
arXiv AI2026-09-17提出中
Workspace Models: Lightweight Robotic Memory via Saliency-Driven Supervision
TL;DRComplex robotic manipulation tasks frequently require a long-term memory of past events an…
arXiv AI2026-09-17提出中
FAMOS: Feed-Forward 3D Articulation Modeling from Sparse Observations
TL;DRModeling articulated objects from sparse monocular views is challenging because each obser…
arXiv AI2026-09-17提出中
Paint-Anything: Unified Any-Color Control for Image Generation and Editing
TL;DRProfessional design requires any-color control: the ability to specify an object's target…
arXiv AI2026-09-17提出中
ERCPMP-Gx: Endoscopic Image and Video Dataset for Morphological, Histopathological, and Genomic Characterization of Colorectal Polyposis
TL;DRHereditary polyposis syndromes can be precursor lesions to colorectal cancer and are assoc…
arXiv AI2026-09-17提出中
Quantifying Overclaiming Propensity in Frontier LLM Agents
TL;DRFrontier coding agents are increasingly trusted to work autonomously for long periods, yet…
arXiv AI2026-09-17提出中
Unifying Models of Intergroup Hostility in Online Discourse
TL;DRHostile rhetoric toward social groups can normalize exclusion and justify mistreatment, as…
arXiv AI2026-09-17提出中
An Empirical Study of Harness Design for Coding Agents
TL;DRCoding harnesses shape how autonomous coding agents translate model capabilities into long…
arXiv AI2026-09-17提出中
JEPA-Anything: Learning Predictive Models across Different Worlds
TL;DRWorld modeling enables intelligence to anticipate consequences, guide interventions, and l…
arXiv AI2026-09-17提出中
RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning
TL;DRMulti-turn agents trained with reinforcement learning (RL) receive a single scalar reward…
arXiv AI2026-09-17提出中
Harm Laundering in GPT Models: Evidence That Gender Discrimination Is Transformed Rather Than Reduced Across Safety-Trained Generations
TL;DRSafety evaluations for large language models rely on surface-form classifiers that report…
arXiv AI2026-09-17提出中
GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies
TL;DRAction chunking is widely used for action generation and execution in Vision-Language-Acti…
arXiv AI2026-09-17提出中
Semantic Action Graph: A Shared Representation for Agent Grounding and Human Interpretation of Sports Highlights
TL;DRGenerative agents are increasingly used to select and narrate video highlights, but they t…
arXiv AI2026-09-17提出中
Prediction-Powered Smoothing and Validation for Disaggregated AI Evaluation
TL;DREvaluating an AI system requires disaggregated assessment, as performance varies across do…
arXiv AI2026-09-17提出中
RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents
TL;DREffective troubleshooting agents in enterprise customer support depend on retrieving actio…
arXiv AI2026-09-17提出中
Large Language Models as Falsifiers for Cyber-Physical Systems
TL;DRFalsification searches for counterexamples to formal specifications in cyber-physical syst…
arXiv AI2026-09-17提出中
dQwen3.5: Hybrid-Attention Diffusion Language Models
TL;DRAdapting a pretrained autoregressive (AR) model is a cost-efficient route to a diffusion l…
arXiv AI2026-09-17提出中
On-Demand Attention: Language Models Know When to Recall
TL;DRReasoning and agentic workloads increasingly demand efficient long-context inference. Yet…
arXiv AI2026-09-17提出中
Q&A on Any Spreadsheet Requires Interpreting Its Grid Structure
TL;DRSemantic cell annotation improves chunking interpretability for spreadsheets in LLM-driven…
OpenAI2026-09-16新兴中
Helping older adults use AI in everyday life
TL;DRHelping older adults use AI in everyday life
OpenAI2026-09-16新兴中
Reimagining advertising with AI
TL;DRReimagining advertising with AI
OpenAI2026-09-16新兴中
How to connect AI usage to business value
TL;DRHow to connect AI usage to business value
OpenAI2026-09-16新兴中
Our framework for reporting model misalignment
TL;DROur framework for reporting model misalignment
OpenAI2026-09-16新兴中
How workers are unlocking new ways of working
TL;DRHow workers are unlocking new ways of working
arXiv AI2026-09-16提出中
Objective vs. Search: Decomposing What Makes a Good Tokeniser
TL;DRTwo dominant tokenisation algorithms are used by modern language models: byte-pair encodin…
arXiv AI2026-09-16提出中
A Zeroth-Order Paradigm for LLM Preference Alignment
TL;DRDirect preference alignment methods are widely used to align large language models (LLMs)…
arXiv AI2026-09-16提出中
PANORAMA: Panoptic Grounded Captioning via Mask Proposal Selection
TL;DRIntelligent systems that act in the world require image understanding that is both compreh…
arXiv AI2026-09-16提出中
Dreaming the Sound of Contact: Leveraging Video and Audio Generation for Zero-Shot Force-Aware Manipulation and Data Generation
TL;DRRecent advances in video generation allow robots to learn manipulation trajectories from g…
arXiv AI2026-09-16提出中
ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments
TL;DRScientific code repositories encode decades of human knowledge in executable models, metho…
arXiv AI2026-09-16提出中
Cognitive Extensions for Dual-Process Language Agents: Memory and Self-Reflection in Interactive Environments
TL;DRLanguage agents remain brittle in interactive environments, where success requires long-ho…
arXiv AI2026-09-16提出中
Affora: A Design System for Agent-Friendly Interfaces
TL;DRComputer-use agents increasingly operate software designed for people, but interfaces ofte…
arXiv AI2026-09-16提出中
Flag Game: A Toy Model for Mechanistic Swarm Interpretability
TL;DREmergent coordinated behaviors of AI agents are starting to present critical safety risks.…
arXiv AI2026-09-16提出中
Playing log(N)-Questions over Wikipedia Abstracts: Communication Efficiency Between Paired Frontier Models
TL;DRWe evaluate six frontier language models on the two-agent $\log(N)$-Questions game. A ques…
arXiv AI2026-09-16提出中
rMuscle: Robotic Muscle Memory for Efficient Vision-Language-Action Model Inference
TL;DRFactory work is a promising early scenario for embodied AI: assigning repetitive manual jo…
arXiv AI2026-09-16提出中
Monitoring and Discovering Reward Hacking with Internal Representations during LLM Evaluations
TL;DRAs models scale, reward hacking becomes more frequent, more sophisticated, and more conseq…
arXiv AI2026-09-16提出中
Prepared Or Unprepared? Evaluating Healthcare Workforce Readiness for Clinical Adoption of Artificial Intelligence in Nigeria
TL;DRArtificial intelligence (AI) is increasingly integrated into healthcare systems worldwide,…
arXiv AI2026-09-16提出中
Reporting Practice Matters: The Impact of Reference Choice on Chest X-ray Report Evaluation
TL;DRRadiologists follow heterogeneous reporting practices. Two radiologists examining the same…
arXiv AI2026-09-16提出中
Securing quantum error correction against misleading advice from AI agents
TL;DRCan an attacker turn influence over an artificial intelligence (AI) adviser into a harmful…
arXiv AI2026-09-16提出中
MUSE: Benchmarking Large Vision-Language Models on Multi-Modal Understanding in Situated Education
TL;DRLarge vision-language models have achieved remarkable progress in multi-modal understandin…
arXiv AI2026-09-16提出中
Probabilistic Linear Explanations
TL;DRFormal explainability provides mathematically grounded justifications for individual predi…
arXiv AI2026-09-16提出中
Double descent is the principle of least action
TL;DRThe test error of a model plotted against its number of parameters $d$ falls, peaks when t…
arXiv AI2026-09-16提出中
RLLBC-Lib: An Educational Code Library for Reinforcement Learning and Learning-Based Control
TL;DRReinforcement learning (RL) is an exciting concept as well as a remarkable success story w…
arXiv AI2026-09-16提出中
Long-Lived Characters, Local Inference: Incremental Memory Maintenance for Game NPCs
TL;DRA game character should not have to reread its entire life before every conversation. For…
arXiv AI2026-09-16提出中
Social Laws for Multi-agent Coordination in Stochastic Environments
TL;DRIn multi-agent environments, coordinating agents to prevent interference and ensure robust…
OpenAI2026-09-16新兴中
Hex turns complex analysis into visual reports with GPT‑6 Astra
TL;DRHex turns complex analysis into visual reports with GPT‑6 Astra
Google DeepMind2026-09-15新兴中
Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Google AI Blog2026-09-15新兴中
New insights from Google’s AI & Economy ATLAS
TL;DRWe’ve translated ATLAS’s millions of global data points into an interactive, open-access e…
Google AI Blog2026-09-15新兴中
AI for Societal Impact
TL;DRExplore this collection to see how experts and local leaders are using AI breakthroughs to…
Google AI Blog2026-09-15新兴中
Building AI to accelerate science and improve lives
TL;DRThe true measure of AI is who it helps. Here’s how it’s impacting lives today. We're focus…
Google AI Blog2026-09-15新兴中
AI for everyone in every language
TL;DRWe’re moving beyond traditional text translation to build models that understand the world…
Hugging Face2026-09-15新兴中
Your Agent Aced the Task. Will It Do It Again?
TL;DRThat is embarrassing onstage. In production, it is a reliability problem: a workflow that…
Google Gemini2026-09-15新兴中
Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking
TL;DRGemini 3.8 Live and Gemini 3.8 Live Extended Thinking are our most advanced live dialogue…
Google Gemini2026-09-15新兴中
4 ways to tackle household chores with Gemini
TL;DRHere’s how Gemini can help you with household chores, like meal planning and fixing applia…
arXiv AI2026-09-15提出中
Agentic Societies Need a Social Harness
TL;DRAn agentic society is a collection of AI agents that coordinate autonomously across trust…
arXiv AI2026-09-15提出中
ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents
TL;DRWe introduce and release ScienceBuddy, an interactive scientific research workspace that b…
arXiv AI2026-09-15提出中
PhysStream: Streaming Physics-Grounded Video Generation with Structured Scene Memory and Fine-Grained Motion Control
TL;DRInteractive control for video generation is moving from coarse prompts toward fine-grained…
arXiv AI2026-09-15提出中
When Should LLMs Abstain? Chain-of-Self-Questioning for Selective Risk Control
TL;DRLarge language models can produce fluent answers when their factual support is weak. This…
arXiv AI2026-09-15提出中
What Breaks Under Pruning in Smart Homes, and When? Evaluating LLM Degradation Across Architectures and Task Complexity
TL;DRPruning can reduce the deployment cost of large language models (LLMs), but its impact on…
arXiv AI2026-09-15提出中
LACE: Layer-Wise Compression for Dynamic Frame Rate Codecs
TL;DRNeural audio codecs are a key component in speech language modeling. However, their high f…
arXiv AI2026-09-15提出中
ENCP: Episode-Normalized Conformal Prediction for Vision-and-Language Navigation
TL;DRUncertainty estimation for Vision-Language-Navigation (VLN) models is a critical task sinc…
arXiv AI2026-09-15提出中
Verifiable Social Reasoning for LLM Assistants
TL;DRLLM assistants are widely used for daily social advice, yet evaluating their social reason…
arXiv AI2026-09-15提出中
LimiX-2: A Contextual Mechanism Network Towards General Structured-Data Intelligence
TL;DRWe introduce LimiX-2, a new model in the LimiX family, developed through model and data sc…
arXiv AI2026-09-15提出中
Det-LIME: Detector-Aware, Multi-Instance Local Interpretable Model-Agnostic Explanations for Automated Marine Mammal Detection
TL;DRDespite the rapid uptake of black-box object detectors in marine mammal research and monit…
arXiv AI2026-09-15提出中
JustFit: 200K-Token LLM Serving on a 24 GiB Laptop with Just-in-Time State Management
TL;DRCapable open-weight models make local coding and reasoning attractive, but their context a…
arXiv AI2026-09-15提出中
Coupled Calibration and Learning: Mitigating Teacher Bias in LLM Distillation without Target-Domain Reward Feedback
TL;DRLarge language model (LLM) distillation aims to transfer the capabilities of a powerful te…
arXiv AI2026-09-15提出中
Decomposition Buys Integrity, Not Yield
TL;DRMulti-agent systems split a task across a tree of agents and justify the split with folklo…
arXiv AI2026-09-15提出中
Evaluating Verified Autonomy in Quantum Engineering
TL;DRReliable quantum engineering is essential for turning quantum phenomena into practical tec…
arXiv AI2026-09-15提出中
Right Tool, Right Job: Native-Language Evaluation, Tokenizer Sensitivity, and Methodological Findings from a French-Only BabyLM
TL;DRWe submit MéTRON-FR, a 125M GPT-2 pretrained on 92.47M words of French, to the BabyLM 2026…
arXiv AI2026-09-15提出中
CareMirror: Bringing Caregiver Wellbeing into the Dementia Care Ecosystem
TL;DRFamily caregivers of people living with dementia shoulder emotional and practical responsi…
arXiv AI2026-09-15提出中
Learning-Guided Planning in Large Dynamic Action Spaces: Budgeted Tree Search for One-to-Many Mobile Charging
TL;DRMany learned sequential decision systems map the current state directly to an action. That…
arXiv AI2026-09-15提出中
Tracking the Unseen: An Occlusion-Robust Framework for Target Tracking Under Full and Long-Term Occlusion
TL;DRReal-time multi-object tracking systems remain highly vulnerable to full and long-term occ…
arXiv AI2026-09-15提出中
Enhancing Accessibility of Medical Texts through Large Language Model-Driven Plain Language Adaptation
TL;DRThis paper addresses the challenge of making complex healthcare information more accessibl…
arXiv AI2026-09-15提出中
Coding Agents Have Converged: Why the SWE-bench Leaderboard Can No Longer Order Its Top Entries, and What to Measure Instead
TL;DRSmall differences on coding-agent leaderboards are often read as an ordering of systems. W…
Google AI Blog2026-09-14新兴中
Watch astronaut Christina Koch and Google’s James Manyika discuss space, technology, and discovery.
TL;DRChristina Koch sits down with James Manyika, Google’s Senior Vice President of Research, L…
Google AI Blog2026-09-14新兴中
DevFest is back
TL;DRDevFest 2026 is back and here’s how you can connect with one of the more than 800 global e…
OpenAI2026-09-14新兴中
Perplexity trusts GPT-6 Astra with end-to-end systems
TL;DRPerplexity trusts GPT-6 Astra with end-to-end systems
OpenAI2026-09-14新兴中
How Fyxer built an AI executive assistant people trust
TL;DRHow Fyxer built an AI executive assistant people trust
Anthropic2026-09-11新兴中
Introducing Claude Corps
TL;DRClaude Corps is a national fellowship for early-career people passionate about extending t…
OpenAI2026-09-11新兴中
Rapidly scaling online storage to serve over 1 billion ChatGPT users
TL;DRRapidly scaling online storage to serve over 1 billion ChatGPT users
OpenAI2026-09-11新兴中
Cognition helps Devin test its own work with GPT‑6 Astra
TL;DRCognition helps Devin test its own work with GPT‑6 Astra
Meta Muse2026-09-10新兴中
Introducing Muse: The World's First Personal AI Agent Built for Everyone
TL;DRMeta 发布面向所有人的个人 AI 智能体 Muse：以 Muse Spark 模型为底座，运行在专门的 Muse Secure VM 中（含独立浏览器），能拆解目标、检索信息…
Anthropic2026-09-10新兴中
Developing nuclear safeguards for AI through public-private partnership
TL;DRWith the NNSA and DOE national labs, we co-developed a classifier that distinguishes conce…
Anthropic2026-09-10新兴中
Higher education advisory board and AI Fluency courses
TL;DRAnthropic announces a Higher Education Advisory Board chaired by Rick Levin and three Crea…
Anthropic2026-09-10新兴中
National Security and Public Sector Advisory Council
TL;DREleven former senators and senior defense, intelligence, and energy officials will help id…
Anthropic2026-09-10新兴中
Updates to Consumer Terms and Privacy Policy
TL;DRClaude Free, Pro, and Max users can now opt in to having conversations used to improve mod…
Anthropic2026-09-10新兴中
Anthropic raises $13B Series F at $183B valuation
TL;DRAnthropic has completed a $13 billion Series F led by ICONIQ, co-led by Fidelity and Light…
Anthropic2026-09-10新兴中
Updating sales restrictions for unsupported regions
TL;DRAnthropic is tightening restrictions to bar entities majority-owned by companies headquart…
Anthropic2026-09-10新兴中
Anthropic joins White House AI education pledge
TL;DRAnthropic signs White House pledge investing in AI education for America's youth, supporti…
Anthropic2026-09-10新兴中
Strengthening safeguards with US CAISI and UK AISI
TL;DRGovernment red-teamers tested Constitutional Classifiers on Claude Opus 4 and 4.1, uncover…
Anthropic2026-09-10新兴中
Chris Ciauri named Managing Director of International
TL;DRChris Ciauri joins Anthropic as Managing Director of International, adding to our global l…
Anthropic2026-09-10新兴中
Deloitte brings Claude to 470,000 people
TL;DRDeloitte will make Claude available across its global network, Anthropic's largest enterpr…
Anthropic2026-09-10新兴中
Rahul Patil joins as Chief Technology Officer
TL;DRRahul Patil brings 20+ years of infrastructure experience from Stripe, AWS, Microsoft, and…
Google AI Blog2026-09-10新兴中
3 ways to prep for your next big race with Search
TL;DRSearch can help runners get race-day ready with registration alerts, tailored training pla…
Hugging Face2026-09-10新兴中
Rebuilding AUTOMATIC1111 with Gradio Workflow
TL;DRWorkflow1111 is a graph of eleven media pipelines built using seventy-three nodes. It brin…
Hugging Face2026-09-10新兴中
Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL
TL;DRLoRA support recently landed in TRL's AsyncGRPOTrainer with PR #7017, and ships with TRL v…
DeepSeek2026-09-10新兴中
DeepSeek-V4.1-Flash: Smarter, Faster, More Efficient
TL;DRDeepSeek-V4.1-Flash: Smarter, Faster, More Efficient
Google Gemini2026-09-10新兴中
The Gemini app is now available for Windows
TL;DRWe’re launching the Gemini app for Windows, the new desktop app built to work alongside yo…
OpenAI2026-09-10新兴中
How a researcher uses Codex and ChatGPT to search for new antimicrobial molecules
TL;DRHow a researcher uses Codex and ChatGPT to search for new antimicrobial molecule…
OpenAI2026-09-10新兴中
Now everyone can put data to work
TL;DRNow everyone can put data to work
OpenAI2026-09-10新兴中
Introducing ChatGPT for Financial Services
TL;DRIntroducing ChatGPT for Financial Services
OpenAI2026-09-10新兴中
Expanding AI access and cyber defense for federal, state, local, and tribal governments
TL;DRExpanding AI access and cyber defense for federal, state, local, and tribal gove…
OpenAI2026-09-10新兴中
Build more natural voice experiences with GPT‑Live‑1 in the API
TL;DRBuild more natural voice experiences with GPT‑Live‑1 in the API
OpenAI2026-09-10新兴中
Introducing the Agents API
TL;DRIntroducing the Agents API
Anthropic2026-09-10新兴中
Anthropic expands to India with Bengaluru office
TL;DRAnthropic will open a Bengaluru office in early 2026, its second APAC location, and improv…
Anthropic2026-09-10新兴中
Salesforce and Anthropic expand partnership
TL;DRClaude becomes a preferred model in Salesforce's Agentforce for regulated industries.
arXiv AI2026-09-10提出中
GPU-CFR: 80x Faster Counterfactual Regret Minimization by Compiling the Game to Static Dataflow and CUDA Graph Replay
TL;DRCounterfactual regret minimization (CFR) is one of the few large numerical workloads that…
arXiv AI2026-09-10提出中
General Quantification of Covariate and Concept Shifts
TL;DRGeneralization under distribution shift remains a core challenge in modern machine learnin…
arXiv AI2026-09-10提出中
Data Scarcity and Model Sparsity: Mixtures-of-Experts Overfit More to Repeated Data
TL;DRAs the supply of human-written text is exhausted, it has become standard practice to repea…
arXiv AI2026-09-10提出中
Can Edge-Deployable Vision-Language Models Identify Species?
TL;DRCamera traps often run in the field on edge hardware with limited or no connectivity, maki…
arXiv AI2026-09-10提出中
Generative Marketing Mix Modeling: A Causal Inference Framework Linking GEO and GEM to Business Impact
TL;DRGenerative artificial intelligence changes how firms reach customers, but standard marketi…
arXiv AI2026-09-10提出中
Distance generalization in transformers: why bother with positional encoding?
TL;DROut-of-distribution length generalization, namely to extrapolate a task from short to long…
arXiv AI2026-09-10提出中
Artificial Id: Drive and Persistent Alignment in Agentic AI
TL;DRAgentic AI is moving from bounded task execution toward systems that retain consequential…
arXiv AI2026-09-10提出中
MindTopo: Can Foundation Models Reason in Topological Space?
TL;DRSpatial reasoning depends not only on metric properties such as distance, angle, and shape…
arXiv AI2026-09-10提出中
Nuha-Speech: Building General-Purpose Arabic Speech-LLMs
TL;DRAs Speech Large Language Models (speech-LLMs) become increasingly multilingual, Arabic rem…
arXiv AI2026-09-10提出中
Domain-Specific Hallucination Detection in Large Language Models
TL;DRLarge language models generate fluent text that can contain unfaithful claims -- a phenome…
arXiv AI2026-09-10提出中
Biology-in-the-loop: Amortized Adaptive Hit Discovery in CRISPR Screens
TL;DRMany biological discovery problems require experiments to be selected sequentially under c…
arXiv AI2026-09-10提出中
On the Regularization Landscape for the Linear Recommendation Models
TL;DRRecently, a wide range of recommendation algorithms inspired by deep learning techniques h…
arXiv AI2026-09-10提出中
The Last AI Built by Humans: Toward Genuine Recursive Self-Improvement
TL;DRRecursive self-improvement (RSI) enables AI systems to turn experience and feedback into p…
arXiv AI2026-09-10提出中
Augustinian BabyLM: What Ostensive Definition Can and Cannot Teach a Small Language Model
TL;DRA language model normally begins training with random word embeddings: whatever 'banana' m…
arXiv AI2026-09-10提出中
Epistemic orientation predicts legislative effectiveness among members of the US Congress
TL;DRTruth and evidence-based communication provide important foundations for democratic govern…
arXiv AI2026-09-10提出中
RetroThinker: Enabling Retrospective Thinking in Speech LLMs
TL;DRSpeech large language models (SpeechLLMs) offer reduced latency and retain paralinguistic…
arXiv AI2026-09-10提出中
Explainability Assistant: A Conversational XAI Interface for Interpreting Energy Consumption Models
TL;DREnergy consumption forecasting relies on increasingly complex machine learning (ML) models…
arXiv AI2026-09-10提出中
From Parameters to Answers: How LLMs Retrieve and Use Their Internal Knowledge
TL;DRHow does a language model's dependence on query-routing information and target knowledge c…
arXiv AI2026-09-10提出中
IndicTriMix: Developing Language Identification Datasets and Models for Tri-Language Code-Mixing
TL;DRLanguage identification in code-mixed text, largely observed in social media, is highly es…
arXiv AI2026-09-10提出中
Model-Aware Schedules Improve Generation via Fiberwise Optimal Transport
TL;DRDiffusion and flow-matching schedules control the signal and noise coefficients that mix d…
Anthropic2026-09-10新兴中
Detecting and countering misuse of AI: August 2025
TL;DRAnthropic's August 2025 threat intelligence report on AI cybercrime and other abuses.
Google AI Blog2026-09-09新兴中
Get ready for the game with new football features in Search
TL;DRTrack live game feeds, explore detailed stats, and get custom fantasy recommendations dire…
Google AI Blog2026-09-09新兴中
Recreating a 70-year love story frame by frame
TL;DRDiscover how filmmakers and Google DeepMind used AI to recreate a couple's unrecorded past…
Hugging Face2026-09-09新兴中
IBM releases SOTA Granite Time Series PatchTST-FM-r2 model with commercial-friendly license
TL;DRTime-series foundation models are changing the way forecasting systems are built. Instead…
Google Gemini2026-09-09新兴中
4 ways Gemini makes administrative chores quick and easy
TL;DRPeople are turning to Gemini for help with complex government tasks, like filing taxes, pr…
OpenAI2026-09-09新兴中
The AI policy window is open. We need to act.
TL;DRThe AI policy window is open. We need to act.
OpenAI2026-09-09新兴中
GPT-6 Astra: The next generation in intelligence for work
TL;DRGPT-6 Astra: The next generation in intelligence for work
OpenAI2026-09-09新兴中
Paul Christiano joins OpenAI Foundation Board
TL;DRPaul Christiano joins OpenAI Foundation Board
Anthropic2026-09-09新兴中
Our compliance framework for California's SB 53
TL;DRAnthropic's Frontier Compliance Framework discloses how we assess and manage catastrophic…
Anthropic2026-09-09新兴中
Introducing Labs
TL;DRAnthropic Labs incubates experimental AI products. Mike Krieger joins Ben Mann on Labs, wh…
Anthropic2026-09-09新兴中
Irina Ghose named Managing Director of India
TL;DRFormer Microsoft India MD Irina Ghose will lead Anthropic's first Bengaluru office, focusi…
Anthropic2026-09-09新兴中
Mariano-Florentino Cuéllar appointed to Long-Term Benefit Trust
TL;DRFormer California Supreme Court Justice Mariano-Florentino Cuéllar joins Anthropic's Long-…
Anthropic2026-09-09新兴中
Anthropic partners with the UK Government on GOV.UK
TL;DRThe UK selected Anthropic to build and pilot an AI assistant for GOV.UK, starting with car…
Anthropic2026-09-09新兴中
ServiceNow chooses Claude
TL;DRServiceNow selects Claude as the default model for its Build Agent and rolls Claude out to…
Anthropic2026-09-09新兴中
Apple’s Xcode now supports the Claude Agent SDK
TL;DRXcode 26.3 integrates the Claude Agent SDK, letting developers use Claude for autonomous,…
Anthropic2026-09-09新兴中
Claude is a space to think
TL;DRClaude will remain ad-free. We explain why advertising incentives are incompatible with a…
Anthropic2026-09-09新兴中
Claude Opus 4.6
TL;DRWe're upgrading our smartest model. Across agentic coding, computer use, tool use, search,…
Anthropic2026-09-09新兴中
Covering electricity price increases
TL;DRAnthropic commits to covering consumer electricity price increases caused by its data cent…
Anthropic2026-09-09新兴中
Anthropic is donating $20 million to Public First Action
TL;DRAnthropic is donating $20 million to Public First Action, a 501(c)(4) focused on AI issues…
Anthropic2026-09-09新兴中
Anthropic raises $30B Series G at $380B valuation
TL;DRAnthropic raised $30 billion, led by GIC and Coatue, with run-rate revenue of $14 billion…
arXiv AI2026-09-09提出中
IdeaAMBIG: Benchmarking Implementation-Critical Gaps in Research-Idea Specifications
TL;DRA research idea may be novel, coherent, and scientifically plausible, yet its proposed met…
arXiv AI2026-09-09提出中
Show-Harness: Just a VLM Agent Can Play Robots
TL;DRFoundation vision-language models (VLMs) exhibit broad intelligence about the world, yet t…
arXiv AI2026-09-09提出中
IBIB: A Protocol for Measuring Enterprise AI Systems by Serving Route, Not Model Identifier
TL;DREnterprises deploy systems, not checkpoints. Usable capability depends jointly on weights,…
arXiv AI2026-09-09提出中
Semigroup-JEPA: Latent Dynamics Consistency for Zero-Shot Physics Generalization
TL;DRJoint-Embedding Predictive Architecture (JEPA) world models learn a compact latent represe…
arXiv AI2026-09-09提出中
JarvisGUI: Towards Cross-Device GUI Agents with Dynamic Task Composition
TL;DRReal-world GUI usage frequently involves workflows that span multiple devices and platform…
arXiv AI2026-09-09提出中
Building Multilingual Bridges: Data Mixing as the Pillar of Generalization for In-Language Reasoning
TL;DRReasoning language models have made substantial advances on a variety of complex tasks, ye…
arXiv AI2026-09-09提出中
ConvMem: Convolutional Memory for Long-Context Reasoning
TL;DRWhile Large Language Models (LLMs) have demonstrated impressive capabilities, they often s…
arXiv AI2026-09-09提出中
Forgetting Only What Matters: Layer-Selective Unlearning toward Robust LLMs
TL;DRLarge Language Models (LLMs) can memorize and reproduce sensitive, copyrighted, or otherwi…
arXiv AI2026-09-09提出中
Do speech foundation models really learn words?
TL;DRSelf-supervised speech foundation models are now used in a wide array of downstream applic…
arXiv AI2026-09-09提出中
Emergency Department Revisit Quality Review Screening: Exploring Human Decision-Making and Artificial Intelligence Support
TL;DRBackground: Emergency Department (ED) return visits are commonly reviewed for quality assu…
arXiv AI2026-09-09提出中
Fortunate Recall: Ontology-Driven Memory Lifecycle Management for Persistent Coherence in LLMs
TL;DRCurrent LLM memory systems treat all personal facts identically, so stores grow without bo…
arXiv AI2026-09-09提出中
Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization
TL;DRThe growing complexity of content moderation policies presents a critical challenge for th…
arXiv AI2026-09-09提出中
Retrofitting Code Using LLMs to Support Exceptional Behavior
TL;DRException Related Code (ERC), which includes throw statements, conditions (if statements)…
arXiv AI2026-09-09提出中
Rosetta at AlexandriaX-2026: LoRA-Adapted NileChat for Context-Aware Dialectal Arabic Dialogue Translation
TL;DRThis paper describes the Rosetta system for Subtask 1 (Context-Aware English-to-Dialectal…
arXiv AI2026-09-09提出中
MOONWALK: Mediating Operations with Intent-Evidence-Action Alignment Across Junior-Supervisor Review Workflows in Animation/VFX Pre-Production
TL;DRAnimation and VFX pre-production review requires teams to translate loosely specified crea…
arXiv AI2026-09-09提出中
PACE: Perceived-Latency-Aware Cascading Service Routing and Filler Control for QoE-Efficient Retrieval-Augmented Dialogue Serving
TL;DRWe present the PACE, a framework for retrieval-augmented dialogue serving that formalizes…
arXiv AI2026-09-09提出中
OmniMed-FL: A Robust Multimodal Federated Learning Framework for Clinical Diagnosis
TL;DRSimultaneous assessment of medical imaging and patient records is often required in clinic…
arXiv AI2026-09-09提出中
Why Is Video Still So Expensive? A Survey of Inference-Efficiency Mechanisms in Video and Audiovisual LLMs
TL;DRVideo understanding has rapidly evolved toward video large language models (VideoLLMs): sy…
arXiv AI2026-09-09提出中
Cyber-Financial Contagion: Modeling the Propagation of an AI Vendor Compromise Through the Banking System
TL;DRThe banking system now depends on a small set of shared artificial intelligence vendors fo…
arXiv AI2026-09-09提出中
Beyond One-Size-Fits-All: Sample-Adaptive Strategy Routing for Vision Token Pruning in MLLMs
TL;DRMultimodal large language models (MLLMs) process hundreds or thousands of visual tokens pe…
Google DeepMind2026-09-08新兴中
AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome
TL;DRAlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants acros…
Hugging Face2026-09-08新兴中
Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic
TL;DRReal deployments rarely fit the topic-level picture. The same base model may be adapted fo…
Anthropic2026-09-08新兴中
Detecting and preventing distillation attacks
TL;DRAnthropic is an AI safety and research company that's working to build reliable, interpret…
Anthropic2026-09-08新兴中
What we learned mapping a year’s worth of AI-enabled cyber threats
TL;DRAs AI transforms the nature of and methods behind cyberattacks, how well do the techniques…
Anthropic2026-09-08新兴中
Detecting and countering malicious uses of Claude
TL;DRDetecting and Countering Malicious Uses of Claude
Anthropic2026-09-08新兴中
Disrupting the first reported AI-orchestrated cyber espionage campaign
TL;DRA report describing an a highly sophisticated AI-led cyberattack
OpenAI2026-09-08新兴中
How GPT-5.6 Sol helps run quantum computing experiments
TL;DRHow GPT-5.6 Sol helps run quantum computing experiments
OpenAI2026-09-08新兴中
The Work Now Within Reach
TL;DRThe Work Now Within Reach
OpenAI2026-09-08新兴中
Introducing ChatGPT Images 2.5
TL;DRIntroducing ChatGPT Images 2.5
OpenAI2026-09-08新兴中
On the Navier–Stokes Millennium Prize Problem
TL;DROn the Navier–Stokes Millennium Prize Problem
OpenAI2026-09-08新兴中
Funding grants for new research into AI and teen development
TL;DRFunding grants for new research into AI and teen development
OpenAI2026-09-08新兴中
OpenAI expands initiatives to support journalism from classrooms to newsrooms
TL;DROpenAI expands initiatives to support journalism from classrooms to newsrooms
OpenAI2026-09-08新兴中
1Password increases engineering productivity 21% with Codex
TL;DR1Password increases engineering productivity 21% with Codex
arXiv AI2026-09-08提出中
TANGO: Humanoid Navigation in Cluttered Environments with a Whole-Body Vision-Language-Action Model
TL;DRWe study the problem of navigating cluttered indoor environments with a humanoid robot. Un…
arXiv AI2026-09-08提出中
Learning Length-Extrapolatable Recurrent Models
TL;DRRecurrent models provide a natural path to long-context modeling, yet models trained with…
arXiv AI2026-09-08提出中
ReCite: Agentic Reasoning for Faithful Citation
TL;DRAccurate citations are the foundation of academic writing, tracing intellectual origins an…
arXiv AI2026-09-08提出中
Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
TL;DRLarge language models are increasingly deployed as agents that plan over long horizons and…
arXiv AI2026-09-08提出中
Copying explains the collective behavior of AI agents in the wild
TL;DRIn June 2026, thousands of AI agents found that a small public wiki would accept edits fro…
arXiv AI2026-09-08提出中
Studying Image Tokenizers as Visual Languages in Unified Multimodal Models
TL;DRImage tokenizers define the ``visual language'' of unified multimodal models, yet are comm…
arXiv AI2026-09-08提出中
NOAH: Learning the Full Patient Journey. A Longitudinal Multimodal Time-Aware Model for Representation and Forecasting
TL;DRThe digitization of healthcare has generated vast, longitudinal, and multimodal patient re…
arXiv AI2026-09-08提出中
A Data-Driven Framework for Identifying and Prioritizing RPA Opportunities in Healthcare Processes
TL;DRRobotic Process Automation (RPA) is widely used to reduce administrative burden in United…
arXiv AI2026-09-08提出中
Entropy-Regularized Rank-Masked Policy Optimization for Test-Time Reinforcement Learning in Code Generation
TL;DRExisting methods for test-time reinforcement learning (TTRL) derive rewards from answer-le…
arXiv AI2026-09-08提出中
Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails
TL;DRAgent harnesses (the system prompt, tool set, execution hooks, and context-management scaf…
arXiv AI2026-09-08提出中
ExecCritic: Learn to Test, Test to Improve for Coding Agents
TL;DRExecution feedback can guide coding agents toward correct repository repairs, but only whe…
arXiv AI2026-09-08提出中
A Generalization of Amari's Bayesian Duality
TL;DRAmari's contributions to information geometry and machine learning are well known. Here, w…
arXiv AI2026-09-08提出中
Canonical Color as a Lens into Concept Decodability in Vision Encoders and VLMs
TL;DRVisual encoders construct a representation of the image input for Vision-Language models.…
arXiv AI2026-09-08提出中
DeCAL: Towards Physically-Grounded Dexterous Vision-Language-Action Models via Contact-Aware Latent Co-Imagination
TL;DRDexterous manipulation involves contact-rich and fine-grained interactions with the physic…
arXiv AI2026-09-08提出中
MeClear: Cooperative Game-Theoretic Attribution and Risk-Aware Memory Clearance for Long-Horizon LLM Agents
TL;DRLong horizon Large Language Model (LLM) agents rely on external memory systems to preserve…
arXiv AI2026-09-08提出中
SAEScientist-Bench: Can AI Agents Conduct Autonomous SAE Interpretability Research?
TL;DRWhile research on recursive self-improvement (RSI) has predominantly automated model train…
arXiv AI2026-09-08提出中
The Surprising Effectiveness of Approximate Value Iteration in Self-Play
TL;DRCombining search with function approximation has driven major advances in game-playing pro…
arXiv AI2026-09-08提出中
Measuring LLM Sycophancy under Sustained Multi-Turn Pressure
TL;DRLarge language models (LLMs) may abandon correct positions when users push back, exhibitin…
arXiv AI2026-09-08提出中
It's Not RoPE that Creates Sinks: The Role of Self-Concentration and Value-Non-Mixing in Attention
TL;DRLarge Language Models (LLMs) often exhibit "Attention Sink" (AS) and the accompanying "Mas…
arXiv AI2026-09-08提出中
GoDeep: Annotation-Free Open-Vocabulary 3D Scene Understanding via Language-Space Lifting
TL;DROpen vocabulary 3D semantic segmentation methods typically lift CLIP features into 3D. Thi…
OpenAI2026-09-07新兴中
Supporting independent journalism in Ukraine
TL;DRSupporting independent journalism in Ukraine
OpenAI2026-09-06新兴中
An Alien Mind
TL;DRAn Alien Mind
OpenAI2026-09-06新兴中
Research acceleration: The view inside OpenAI
TL;DRResearch acceleration: The view inside OpenAI
Google Gemini2026-09-04新兴中
Create your best tracks yet with Lyria 3.5 in Gemini.
TL;DRLyria 3.5, our best-sounding music generation model, is now available in the Gemini app an…
arXiv AI2026-09-04提出中
WearableQA: A Benchmark for Health Reasoning over Real-World Wearable Data
TL;DRRecent advances in wearable sensing enable continuous monitoring of physiological and beha…
arXiv AI2026-09-04提出中
Diffusion TV: Experiencing Diffusion Models through Tangible, Embodied Interaction
TL;DRDiffusion TV is an interactive AI art installation that offers a tangible and embodied exp…
Google DeepMind2026-09-03新兴中
Introducing WeatherNext 3, our most advanced and accurate global weather AI model
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Hugging Face2026-09-03新兴中
Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps
TL;DRStructured output is one of the most common real-world tasks for LLMs, yet most benchmarks…
Hugging Face2026-09-03新兴中
Give Your Coding Agents a Memory You Own
TL;DREarlier this year, Software Forgets: Agent Traces Are the Memory made the case that coding…
Meta2026-09-03新兴中
ZGateway: Learnings from Putting a Proxy in Front of ZippyDB
TL;DRWe’re introducing ZGateway, the proxy we are using to unify traffic through ZippyDB, Meta’…
Google DeepMind2026-09-02新兴中
Proactive cyber defense for governments and enterprises
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Google DeepMind2026-09-02新兴中
Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Google AI Blog2026-09-02新兴中
Proactive cyber defense for governments and enterprises
TL;DRThe Fairwind Program is a limited access program for governments and trusted partners to u…
Meta2026-09-02新兴中
An Organizational Second Brain: Building an AI That Learns From Experts
TL;DRWe’ve built an AI agent that acts as a secondary expert for a given domain, making deep sp…
Google Gemini2026-09-02新兴中
Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
TL;DRGemini 3.8 Flash and 3.8 Flash Cyber deliver next-generation intelligence for agentic work…
Google Gemini2026-09-02新兴中
MrBeast partners with Gemini to turn impossibly big ideas into reality
TL;DRGoogle is kicking off a multi-year partnership with Jimmy Donaldson, better known as MrBea…
Google DeepMind2026-09-01新兴中
Introducing agentic video understanding with Gemini
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Google AI Blog2026-09-01新兴中
The latest AI news we announced in August 2026
TL;DRHere are Google’s latest AI updates from August 2026
Google Gemini2026-09-01新兴中
Introducing agentic video understanding with Gemini
TL;DRWe’re launching agentic video understanding across our latest Gemini models for improved a…
Google DeepMind2026-08-27新兴中
Gemini Omni 1.1 Flash lets you build with more control
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
Google DeepMind2026-08-27新兴中
Piloting the world's first double-blind AI evaluations
TL;DRPiloting the world's first double-blind AI evaluations
Google Gemini2026-08-27新兴中
Gemini Omni 1.1 Flash lets you build with more control
TL;DRGemini Omni 1.1 Flash brings a new suite of creative controls and generative video capabil…
Google DeepMind2026-08-26新兴中
Intelligent transcription with Gemini 3.5 Transcribe
TL;DRNow you can get more intelligent speech-to-text transcription with Gemini 3.5 Transcribe.
Google Gemini2026-08-26新兴中
7 ways to kick-start back to school using Gemini in Workspace
TL;DRGet a jump on this semester by using Gemini in Google Workspace to organize your notes, tr…
Meta2026-08-24新兴中
MetaRoCE: A New RDMA Transport Built for AI-Scale Ethernet
TL;DRTraining and serving frontier AI models depends on fast, reliable networks that move data…
Meta2026-08-24新兴中
MTIA 300: Meta’s First Training Chip with Built-in NICs and Communication-Offloading Engines
TL;DRMTIA 300 is the first of Meta’s family of in-house training and inference accelerators opt…
Google DeepMind2026-08-21新兴中
From Atari to EVE Online: Building on 15 Years of AI Research in Games
TL;DRGoogle DeepMind partners with game studios to prototype breakthrough AI gameplay.
DeepSeek2026-08-21新兴中
DeepSeek-V4-Flash-Vision-Exp Release: Multimodal API Now Live
TL;DRDeepSeek-V4-Flash-Vision-Exp Release: Multimodal API Now Live
Google DeepMind2026-08-13新兴中
Introducing Gemini 3.7 Flash
TL;DR"}@property --border-mask-color{inherits:false;initial-value:transparent;syntax:""}@proper…
DeepSeek2026-08-13新兴中
DeepSeek-V4-Pro GA Release
TL;DRDeepSeek-V4-Pro GA Release
Google DeepMind2026-08-12新兴中
Putting sign language AI into users’ hands
TL;DRIntroducing sign-language-to-text (SL2T), our breakthrough model powering new sign languag…
Meta2026-08-12新兴中
How We’re Building Scam Alert on WhatsApp With End-to-End Encryption and Verifiability Guarantees
TL;DRWhatsApp is committed to helping people stay safe while protecting the privacy of their me…
Genspark2026-08-06新兴中
The AI Efficiency Frontier: How Enterprises Scale Agentic AI Without Letting Cost Outrun Value
TL;DRGenspark 联合创始人兼 CTO Kay 的技术长文：企业规模化 Agentic AI 的关键不是押注单一模型，而是按「每次成功任务的成本」移动效率前沿。
Meta2026-08-05新兴中
From User Sequences to Scaling Laws: A Multi-Stage Architecture for Meta’s Ads Ranking
TL;DREvery day, Meta’s recommendation platforms handle billions of user interactions, generatin…
Kimi2026-07-17新兴中
Kimi K3：智能的新前沿
TL;DR月之暗面（Moonshot AI）发布旗舰模型 Kimi K3：2.8 万亿参数级的开源 MoE 模型，基于 Kimi Delta Attention（KDA）混合线性注意力与 A…
DeepSeek2026-04-24新兴中
DeepSeek V4 Preview Release
TL;DRDeepSeek V4 Preview Release
Creao AI2026-04-13新兴中
We Built an Agent Platform. Then the Agents Rebuilt It.
TL;DRCREAO（25 人团队）自述 AI-first 工程实践：99% 的生产代码由 AI 编写。
Meta Muse2026-04-08新兴中
Introducing Muse Spark: MSL's First Model, Purpose-Built to Prioritize People
TL;DRMeta Superintelligence Labs（MSL）发布首个模型 Muse Spark：为 Meta 产品量身打造的多模态推理模型，当前驱动 Meta AI 应用与网站…
Manus2026-01-16新兴中
AI Presentation Tools for Education: Top 5 Tools Every Teacher Should Try
TL;DRDiscover the top 5 AI presentation tools every teacher should try in 2026. Learn how Manus…
Manus2025-12-29新兴中
Manus Joins Meta for Next Era of Innovation
TL;DRManus is joining Meta, and we’ll continue delivering our current services while accelerati…
Manus2025-12-18新兴中
Slides Created with Nano Banana Pro Are Now Editable—A First for AI Presentations
TL;DREdit slides created with Nano Banana Pro on Manus
Manus2025-12-17新兴中
Manus Update: $100M ARR, $125M revenue run-rate
TL;DRManus reaches $100M ARR, becoming the fastest startup to hit this milestone. With $125M to…
Manus2025-12-17新兴中
Manus Academy Launch
TL;DRManus launches AI training academy with Build Club for professionals and businesses to get…
Manus2025-12-15新兴中
Introducing Manus 1.6: Max Performance, Mobile Dev, and Design View
TL;DRIntroducing Manus 1.6!
Manus2025-12-10新兴中
Get Your Manus App to Rank on Search Engines: Introducing Built-in SEO
TL;DRWith the latest update to Manus Web app, new capabilities that make the web apps you build…
Manus2025-12-08新兴中
Introducing Custom Domains: Your App, Your Address
TL;DRYour work deserves a real address. With Custom Domains, you can now connect or purchase a…
Manus2025-12-08新兴中
Google Drive Connector on Manus
TL;DRThe **Google Drive Connector** is now available on Manus—a powerful integration that trans…
Manus2025-12-05新兴中
Manus Slides Just Got More Powerful: Studio-Quality Presentations with Nano Banana Pro
TL;DRManus Slides now features Nano Banana Pro for superior AI-generated presentations. Get stu…
DeepSeek2025-12-01新兴中
DeepSeek-V3.2 Release
TL;DRDeepSeek-V3.2 Release
DeepSeek2025-09-29新兴中
Introducing DeepSeek-V3.2-Exp
TL;DRIntroducing DeepSeek-V3.2-Exp
DeepSeek2025-09-22新兴中
DeepSeek-V3.1-Terminus
TL;DRDeepSeek-V3.1-Terminus
DeepSeek2025-08-21新兴中
DeepSeek-V3.1 Release
TL;DRDeepSeek-V3.1 Release
DeepSeek2025-05-28新兴中
DeepSeek-R1-0528 Release
TL;DRDeepSeek-R1-0528 Release
DeepSeek2025-03-25新兴中
DeepSeek-V3-0324 Release
TL;DRDeepSeek-V3-0324 Release