Papers

• MAI-Thinking-1: Building a Hill-Climbing Machine

  Published on: 2026-06-02 Venue: Technical report (Microsoft AI) 1 author
• Writing Code vs. Shipping Code: Productivity Effects Across Generations of AI Coding Tools

  Massachusetts Institute of Technology, National Bureau of Economic Research (NBER), University of Pennsylvania

  Published on: 2026-05-01 Venue: NBER Working Paper Series, No. 35275 3 authors
• Memento: Teaching LLMs to Manage Their Own Context

  Published on: 2026-04-08 Venue: Microsoft Research (AI Frontiers) Technical Report 10 authors
• The Price Reversal Phenomenon: When Cheaper Reasoning Models End Up Costing More

  Carnegie Mellon University, Stanford University, UC Berkeley

  Published on: 2026-03-25 6 authors
• CyberThreat-Eval: Can Large Language Models Automate Real-World Threat Research?

  Hong Kong University of Science and Technology

  Published on: 2026-03-10 8 authors
• Chow-Liu Ordering for Long-Context Reasoning in Chain-of-Agents

  Published on: 2026-03-10 11 authors
• AutoAdapt: An Automated Domain Adaptation Framework for LLMs

  Published on: 2026-03-09 6 authors
• OSS-CRS: Liberating AIxCC Cyber Reasoning Systems for Real-World Open-Source Security

  Georgia Institute of Technology

  Published on: 2026-03-09 10 authors
• StreamWise: Serving Multi-Modal Generation in Real-Time at Scale

  Massachusetts Institute of Technology

  Published on: 2026-03-06 7 authors
• Lost in Stories: Consistency Bugs in Long Story Generation by LLMs

  Singapore University of Technology and Design

  Published on: 2026-03-06 6 authors
• LUMINA: LLM-Guided GPU Architecture Exploration via Bottleneck Analysis

  Published on: 2026-03-06 5 authors
• OpenFrontier: General Navigation with Visual-Language Grounded Frontiers

  ETH Zurich, University of Bonn

  Published on: 2026-03-05 4 authors
• On the Necessity of Learnable Sheaf Laplacians

  University of Cambridge

  Published on: 2026-03-05 4 authors
• Scaling Agentic Capabilities, Not Context: Efficient Reinforcement Finetuning for Large Toolspaces

  Published on: 2026-03-05 6 authors
• Pretrained Vision-Language-Action Models are Surprisingly Resistant to Forgetting in Continual Learning

  University of Texas

  Published on: 2026-03-04 1 author
• Phi-4-reasoning-vision-15B Technical Report

  Published on: 2026-03-04 1 author
• UniG2U-Bench: Do Unified Models Advance Multimodal Understanding?

  Shanghai Jiao Tong University

  Published on: 2026-03-03 1 author
• Beyond Pixel Histories: World Models with Persistent 3D State

  University of Edinburgh

  Published on: 2026-03-03 1 author
• Modular Memory is the Key to Continual Learning Agents

  University of Bremen

  Published on: 2026-03-02 1 author
• Exploratory Memory-Augmented LLM Agent via Hybrid On- and Off-Policy Optimization

  Korea Advanced Institute of Science & Technology

  Published on: 2026-02-26 1 author
• Exploratory Memory-Augmented LLM Agent via Hybrid On- and Off-Policy Optimization

  Korea Advanced Institute of Science & Technology

  Published on: 2026-02-26 1 author
• WebGym: Scaling Training Environments for Visual Web Agents with Realistic Tasks

  University of Illinois Urbana-Champaign

  Published on: 2026-02-25 1 author
• Experiential Reinforcement Learning

  University of Southern California

  Published on: 2026-02-15 1 author
• WizardLM: Empowering large pre-trained language models to follow complex instructions

  Peking University

  1 author
• Florence: A New Foundation Model for Computer Vision

  1 author
• LLM-in-Sandbox Elicits General Agentic Intelligence

  Tsinghua University

  Published on: 2026-02-12 1 author
• On-Policy Context Distillation for Language Models

  Microsoft Research

  Published on: 2026-02-12 1 author
• CineScene: Implicit 3D as Effective Scene Representation for Cinematic Video Generation

  The University of Hong Kong, Tsinghua University, Zhejiang University

  Published on: 2026-02-06 1 author
• See Less, See Right: Bi-directional Perceptual Shaping For Multimodal Reasoning

  Tsinghua University

  Published on: 2026-02-05 1 author
• LIVE: Long-horizon Interactive Video World Modeling

  The Chinese University of Hong Kong

  Published on: 2026-02-03 1 author
• Closing the Loop: Universal Repository Representation with RPG-Encoder

  University of California San Diego

  Published on: 2026-02-03 1 author
• CUA-Skill: Develop Skills for Computer Using Agent

  Published on: 2026-02-02 1 author
• AgentRx: Diagnosing AI Agent Failures from Execution Trajectories

  University of Illinois Urbana-Champaign

  Published on: 2026-02-02 6 authors
• X-Coder: Advancing Competitive Programming with Fully Synthetic Tasks, Solutions, and Tests

  Tsinghua University, Wuhan University

  Published on: 2026-02-01 1 author
• LLM-42: Enabling Determinism in LLM Inference with Verified Speculation

  Microsoft Research

  Published on: 2026-01-30 1 author
• Lost in Transmission: When and Why LLMs Fail to Reason Globally

  Published on: 2026-01-30 1 author
• Efficient Autoregressive Video Diffusion with Dummy Head

  ETH Zurich, Johns Hopkins University, Tsinghua University, University of Science and Technology of China

  Published on: 2026-01-28 1 author
• Can LLMs Clean Up Your Mess? A Survey of Application-Ready Data Preparation with LLMs

  Hong Kong University of Science and Technology, Massachusetts Institute of Technology, Shanghai Artificial Intelligence Laboratory, Shanghai Jiao Tong University, Tsinghua University

  Published on: 2026-01-22 1 author
• Collaborative Multi-Agent Test-Time Reinforcement Learning for Reasoning

  Massachusetts Institute of Technology

  Published on: 2026-01-15 1 author
• Controlled LLM Training on Spectral Sphere

  Renmin University of China

  Published on: 2026-01-13 1 author
• InfiniteWeb: Scalable Web Environment Synthesis for GUI Agent Training

  Peking University

  Published on: 2026-01-08 1 author
• Thinking with Blueprints: Assisting Vision-Language Models in Spatial Reasoning via Structured Object Representation

  National University of Singapore

  Published on: 2026-01-05 1 author
• From Word to World: Can Large Language Models be Implicit Text-based World Models?

  Southern University of Science and Technology

  Published on: 2025-12-21 1 author
• Sigma-MoE-Tiny Technical Report

  Microsoft Research

  Published on: 2025-12-19 1 author
• FlashPortrait: 6x Faster Infinite Portrait Animation with Adaptive Latent Prediction

  Fudan University

  Published on: 2025-12-18 1 author
• Spatia: Video Generation with Updatable Spatial Memory

  The University of Sydney

  Published on: 2025-12-17 1 author
• Native and Compact Structured Latents for 3D Generation

  Tsinghua University

  Published on: 2025-12-16 1 author
• Wait, Wait, Wait... Why Do Reasoning Models Loop?

  Massachusetts Institute of Technology

  Published on: 2025-12-15 1 author
• Glance: Accelerating Diffusion Models with 1 Sample

  Wissenschaftliche Hochschule für Unternehmensführung

  Published on: 2025-12-11 1 author
• Semantics Lead the Way: Harmonizing Semantic and Texture Modeling with Asynchronous Latent Diffusion

  Beijing Jiaotong University

  Published on: 2025-12-05 1 author