Research Digest Archive

Plain static pages generated from the workspace digest markdown files. This directory is self-contained and can be published as-is.

Lumen Research Digest — 2026-10-09

2026-10-09 · 7 items

• 1. FloorSAV: Elucidating Spatial Audio-Visual Context with 2D Floormap for AV-LLMs
• 2. GATOR: Generative and Agentic 3D Object Reconstruction From Casual Images
• 3. SuperNav: An Agentic Navigation System for Any Task in Any Scene
• 4. SpaceCast-Bench: Evaluating Predictive Spatial Reasoning in Vision-Language Models
• 5. 4-Tensor Attention Model for Semantic Physical Reality
• 6. VideoEvolve: Co-Evolving Memory and Retrieval for Long Video Understanding
• 7. Rendering-Free Lookahead for Question-Guided Active Vision

Lumen Research Digest — 2026-10-08

2026-10-08 · 7 items

• 1. TileSkipper: Region-Adaptive Tile Pruning for 3D Gaussian Splatting
• 2. SPLATIFY: Reproduce, Discover, Innovate! From Papers and Ideas to Trainable 3DGS Code
• 3. DeltaSplat: Iterative Gaussian Refinement for Pose-Free Feed-Forward 3D Gaussian Splatting
• 4. vLLM-Omni Technical Report: A Unified Serving Runtime for Omni-Modality Generation
• 5. Context-aware Attention-based Gaussian Mixture Models for Vehicular Trajectory Prediction
• 6. SearchWorld: Spatial Value-Grounded Imagination for UAV Object Search via World Models
• 7. Humanity's Sixth Sense: Benchmarking Intuitive Visual Reasoning in Multimodal Models

Lumen Research Digest — 2026-10-07

2026-10-07 · 7 items

• 1. MoonGS: High-quality Representation of the Lunar Surface via Gaussian Splatting Using Robust Depth Features from Image Pairs
• 2. OpenSplatGraph: From Dense Semantic Maps to Structured Scene Graphs for Open-Vocabulary Robot Perception
• 3. DepthWorld: 3D World Model for Robot Manipulation
• 4. WorldSolver: Can LLM Agents Simulate the Physical Dynamics via Solver Generation?
• 5. Rethinking Visual Provenance: Detection and Watermarking Across Direct Visual Generation and LLM-Driven Code Rendering
• 6. DensiTok: Making Feed-Forward 3D Gaussian Splatting See More Views Than It Is Given
• 7. Robotizing Human Videos with Physically Consistent Interactions

Lumen Research Digest — 2026-10-06

2026-10-06 · 7 items

• 1. Mobile-4DGS: Unified Static-Dynamic Real-time Mobile Gaussian Splatting
• 2. VideoTapestry: Query-Adaptive Memory Refinement for Multi-Agent Long-Video Understanding
• 3. Have I Scene This Before? Spatially Grounded Conversational Memory for Complex Queries in Egocentric Assistants
• 4. Video2World: Benchmarking Coding Agents for Interactive World Modeling from Embodied Videos
• 5. BrainTRACE: Tracing Longitudinal, Multimodal, and Volumetric Evidence in Brain MRI Clinical Reasoning
• 6. Code2Games: Enabling Coding Agents for Gaming World Generation
• 7. ArticuTable: Generating Instance-Level Interactive Rigid-Articulated 3D Tabletop Scenes from a Single Image

Lumen Research Digest — 2026-10-05

2026-10-05 · 7 items

• 1. ManifoldSplat: Language-Guided Semantic Shape Editing of 3D Gaussian Head Avatars
• 2. EdgeAgent: Orchestrating On-Device LLM inference for End-User Multi-Agent Systems on CPU-GPU Unified Memory Architectures
• 3. HazardWeaver: Scientific Route Selection for Hazard Analysis Agents
• 4. CORNAV: Construction-Aware Reasoning for Robot Navigation on Active Worksites
• 5. Beyond Single Videos: Benchmarking and Active Evidence Seeking for E-Commerce Cross-Video Reasoning
• 6. OmniAct3D: Leveraging Foundation Geometry and Evidence-Grounded Reasoning for Panoramic 3D Detection
• 7. WebFovea: When the Model Is Right but the Click Is Wrong -- Reliable Round Trips for Vision-Based Web Agents on Live Websites

Lumen Research Digest — 2026-10-04

2026-10-04 · 7 items

• 1. SkeleWAM: Skeleton World-Action Modeling for Efficient Robotic Manipulation
• 2. LiteReality-Agent: An Agentic System for Interactable 3D Indoor Scene Reconstruction
• 3. World Observer: Joint Actor-Observer Generation for Persistent World Modeling
• 4. Code Owns the Simulation, Jev Owns the Evaluation
• 5. Continual Learning for 6-DoF Grasp Synthesis via Experience and Demonstrations
• 6. Before It Fades: Reinforcing Temporal Representations at Inference Time in VideoLLMs
• 7. PAGER: Partial-to-global Alignment via Geometric and Relational Distillation

Lumen Research Digest — 2026-10-03

2026-10-03 · 7 items

• 1. Reconstruct, Practice, Go Real: Guided Self-Improvement for Embodied Agents
• 2. CoVisco: Codec-Native Vision Encoder with Native Token Compression for Unified Image-Video Understanding
• 3. TouchTherm: Building Multimodal Digital Twins of Objects for Tactile and Thermal Rendering
• 4. GenCOPE: Syn2Real Generalized Category-Level Object Pose Estimation for Robotic Picking
• 5. MEGA: Object-Level Mesh Extraction from 3D Gaussian Splatting via Spatial Visual Distillation
• 6. Lens Flare Removal and Reconstruction
• 7. FutureWorlds: Learning Robotic World Models from Alternative Futures

Lumen Research Digest — 2026-10-02

2026-10-02 · 7 items

• 1. DuoMind: Enabling Distributed Multi-Robot Coordination with Semantic Communication
• 2. Are Frontier VLM Agents Ready to Be Robot Generalists? An Empirical Study with the Embodied Agent Arena
• 3. MemFit: Efficient Long-Term Agentic Memory
• 4. CtrlWAM: Controllable World Action Models with Aligned Intent and Foresight
• 5. Explore, Execute, Evolve: A Skill Acquisition and Reuse Loop for Embodied Agents
• 6. Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents
• 7. VTR-Bench: A Systematic Benchmark for Evaluating Visual Text Rendering in Video Generation

Lumen Research Digest — 2026-10-01

2026-10-01 · 7 items

• 1. WorldAuditBench: Interactive 3D World Auditing with Multimodal Agents
• 2. MemLife: Curating and Reasoning over Long-Term Egocentric Video Memories
• 3. LongEmo: Towards Emotion Understanding and Reasoning in Long Videos
• 4. EgoTools: Towards Tool-Centric Reasoning in Real-World Egocentric Videos
• 5. STARS: From Spatiotemporal Dynamics to Social Representations in Human-Robot Interaction
• 6. RoboAssist: Interactive Human-Humanoid Planning for Long-Horizon Surgical Assistance
• 7. Breaking Babel: A Self-Evolving Multi-Agent System for Long-Form Subtitle Translation

Lumen Research Digest — 2026-09-30

2026-09-30 · 7 items

• 1. Imagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before Answering
• 2. Exemplar2VQA: A Scalable Exemplar-Driven Visual Question Answering Generation Framework via Multi-Agent Coding
• 3. Code4Scene: Benchmarking Coding Agents for Constructing and Editing 3D Scenes
• 4. DispFlow-GS: Displacement Flow Supervision with Motion Disentangling for Monocular Deformable 3D Gaussian Splatting
• 5. Distilling Privileged Control Barrier Functions into RGB-Only Safety Filters for Dynamic Visual Navigation
• 6. OmniVCBench: Benchmarking Evidence-Grounded Multimodal Reasoning Towards AI Virtual Cells
• 7. Concealing LLM-Based Multi-Agent Topology via Phantom Structure Injection

Lumen Research Digest — 2026-09-29

2026-09-29 · 7 items

• 1. CodeActionBench: Evaluating Agentic Code-as-Policy for Embodied Manipulation
• 2. Long Time No See: Benchmarking VLMs for Out-of-Sight Spatiotemporal Reasoning in Egocentric Videos
• 3. Recent Advances in Agentic Agri-Robotic Phenotyping: A Perspective Review from Fragmented Multimodal Sensing to Unified PhenoAgent Intelligence
• 4. Gaussian Splatting-based Volumetric Video Compression with Sparse 4D Anchors
• 5. Robot-GST: geometry-aware spatial-temporal robot policy representation and evaluation
• 6. JRDB-AVR: An Active Visual Reasoning Benchmark for Embodied Agents in Real-World Environments
• 7. GenNVS: Geometry-enhanced Novel View Synthesis via Disentangled 3D Prior

Lumen Research Digest — 2026-09-28

2026-09-28 · 7 items

• 1. DualManip: Agentic Dynamic Manipulation via Dual-Path Semantic Reasoning and Geometric Adaptation
• 2. InternW0-$\Delta$: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data
• 3. SatNav: A Scalable Benchmark for Long-Horizon UAV Vision-Language Navigation from Satellite Imagery
• 4. Gauss What You Need: Compact Gaussian Splatting Across Scene Scales
• 5. MetaPermit: Scalable and Auditable Access Control for AI Agents via LLM-Inferred Meta-Attributes
• 6. Structured Reasoning Agentic Framework for Interpretable Critical View of Safety Assessment
• 7. RECAST: From Log Replay to Closed-Loop Driving Simulation with View-Complete Actors

Lumen Research Digest — 2026-09-27

2026-09-27 · 7 items

• 1. OceanXL: Large-scale Underwater 3D Gaussian Splatting via Block Partitioning and Adaptive Pruning
• 2. Jev-Mobile: Jev as an Executor for Mobile GUI Agents
• 3. Industrial Anomaly Detection via Defect-Grounded Reasoning in Visual Latent Space
• 4. AdaHVLA: Adaptive Harnesses for Long-Horizon Vision-Language-Action Execution
• 5. When Can Agents Forget Their Reasoning? ICLR for Long-Horizon Agent Context Compression
• 6. PolyUMI: Accessible Visual-Tactile-Audio Data Collection for Object Inference and Manipulation
• 7. SARFusion: Scene-Aware Routing Fusion for Robust Camera-LiDAR 3D Object Detection

Lumen Research Digest — 2026-09-26

2026-09-26 · 7 items

• 1. RAPID: Robot Agentic Programming from Demonstrations
• 2. Multimodal Thinking with Renderable Programs
• 3. RACaP: Agentic Reasoning, Acting, and Coding as Policies for Evolvable Robot Learning
• 4. Markerless Multi-Modal Autonomous Robotic Inspection of Large Space Structures
• 5. SEE Challenge 2026: Event-Guided Brightness Adjustment Across a Broad Illumination Range
• 6. Don't Read the Log: Execution Traces Contaminate Verifiers in Video-Generation Agents
• 7. Pistis Technical Report

Lumen Research Digest — 2026-09-25

2026-09-25 · 7 items

• 1. World Action Agent: Harnessing VLMs for Robot Manipulation via World Action Rehearsal
• 2. Beyond Spatial Benchmarks: From Spatial Reasoning to Navigation
• 3. SWE-PolyVision: Benchmarking Cross-Image Abductive Reasoning for Repository-Level Software Engineering
• 4. From Passive Execution to Active Exploration: Agentic Embodied Manipulation in Realistic Environments
• 5. BaseCamp --- An Agentic AI Framework for Automating DNA Sequencing Data Pipelines
• 6. HarnessPAI: An Evolving Harness for Physical AI
• 7. Looks the Same, Answers Differently: Flip-Direction Steering for Robust Vision-Language Reasoning

Lumen Research Digest — 2026-09-24

2026-09-24 · 7 items

• 1. EnSIMem: Entity-Structured Indexing for Long-Term Agent Memory
• 2. GaussPDE: Graph-Based Partial Differential Equation-Driven Rendering for 3D Gaussian Splatting
• 3. ACTS: A multi-tier benchmark evaluating LLM cipher identification under controlled blind conditions
• 4. EmbodiedMemory-Bench: Benchmarking Embodied Memory for Long-Horizon Embodied Tasks
• 5. HINT-Blimp: Human INTent Inference from Multimodal Cues for Robotic Blimps
• 6. CoRelNav: Collaborative Relational Navigation for Multi-Robot Spatially Constrained Semantic Navigation
• 7. Building Socio-Affective Artificial Intelligence for Interactive Multi-Agent Simulations

Lumen Research Digest — 2026-09-23

2026-09-23 · 4 items

• 1. Parallel cut research time and cost in half with GPT‑6 Astra
• 2. Echoverse: Deep, evolving environments for computer-use agents
• 3. Orchard: An open framework for scalable agentic AI
• 4. Higgsfield AI ships new video features in a day with GPT-6 Astra

Lumen Research Digest — 2026-09-22

2026-09-22 · 4 items

• 1. How V7 gives AI agents institutional memory
• 2. Improving synthesis prediction of small molecules at scale with RetroChimera
• 3. Higgsfield AI ships new video features in a day with GPT-6 Astra
• 4. EvoLib: Turning experience into evolving knowledge

Lumen Research Digest — 2026-09-21

2026-09-21 · 4 items

• 1. Helping older adults use AI in everyday life
• 2. Verifying Rust cryptography in SymCrypt, from standards to code
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Reimagining advertising with AI

Lumen Research Digest — 2026-09-20

2026-09-20 · 4 items

• 1. Our framework for reporting model misalignment
• 2. Echoverse: Deep, evolving environments for computer-use agents
• 3. Orchard: An open framework for scalable agentic AI
• 4. Reimagining advertising with AI

Lumen Research Digest — 2026-09-19

2026-09-19 · 4 items

• 1. Hex turns complex analysis into visual reports with GPT‑6 Astra
• 2. Echoverse: Deep, evolving environments for computer-use agents
• 3. Orchard: An open framework for scalable agentic AI
• 4. Reimagining advertising with AI

Lumen Research Digest — 2026-09-18

2026-09-18 · 4 items

• 1. How Cooley is accelerating IPO work with ChatGPT
• 2. GigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation models
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Reimagining advertising with AI

Lumen Research Digest — 2026-09-17

2026-09-17 · 4 items

• 1. Reimagining advertising with AI
• 2. Echoverse: Deep, evolving environments for computer-use agents
• 3. How to connect AI usage to business value
• 4. Orchard: An open framework for scalable agentic AI

Lumen Research Digest — 2026-09-16

2026-09-16 · 5 items

• 1. PanoGS-SLAM: Panoramic 3D Gaussian Splatting SLAM
• 2. How Fyxer built an AI executive assistant people trust
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. RobResilience: Implementing and Evaluating a Resilience Framework for Cyber-Physical Embodied Systems
• 5. You Shall Not Pass into Ring-0! A User Privacy-Friendly Anti-Cheat Architecture for Personal Computers

Lumen Research Digest — 2026-09-15

2026-09-15 · 4 items

• 1. Introducing ChatGPT for Financial Services
• 2. Echoverse: Deep, evolving environments for computer-use agents
• 3. Orchard: An open framework for scalable agentic AI
• 4. Now everyone can put data to work

Lumen Research Digest — 2026-09-14

2026-09-14 · 4 items

• 1. Expanding AI access and cyber defense for federal, state, local, and tribal governments
• 2. MindTopo reveals VLMs’ spatial reasoning abilities
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Now everyone can put data to work

Lumen Research Digest — 2026-09-13

2026-09-13 · 4 items

• 1. How a researcher uses Codex and ChatGPT to search for new antimicrobial molecules
• 2. Introducing CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement
• 3. Cognition helps Devin test its own work with GPT‑6 Astra
• 4. Echoverse: Deep, evolving environments for computer-use agents

Lumen Research Digest — 2026-09-12

2026-09-12 · 6 items

• 1. Learning Agent-based Model Predictive Control for Holistic Vehicle Performance
• 2. Rapidly scaling online storage to serve over 1 billion ChatGPT users
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. BlueSTAR: Tiered Agentic Architecture for Autonomous Cyber Defense
• 5. Artificial Id: Drive and Persistent Alignment in Agentic AI
• 6. Perplexity trusts GPT-6 Astra with end-to-end systems

Lumen Research Digest — 2026-09-11

2026-09-11 · 6 items

• 1. SenseNova-U1.5: Towards Native Unified Visual Intelligence
• 2. Introducing the Agents API
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. MindTopo: Can Foundation Models Reason in Topological Space?
• 5. Domain-Specific Hallucination Detection in Large Language Models
• 6. Now everyone can put data to work

Lumen Research Digest — 2026-09-10

2026-09-10 · 6 items

• 1. Programmable World Model
• 2. Paul Christiano joins OpenAI Foundation Board
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. JarvisGUI: Towards Cross-Device GUI Agents with Dynamic Task Composition
• 5. Towards Tackling Application Logic Flaws through Autonomous Formal-Logic Modeling and Automated Reasoning
• 6. GPT-6 Astra: The next generation in intelligence for work

Lumen Research Digest — 2026-09-09

2026-09-09 · 6 items

• 1. ReCite: Agentic Reasoning for Faithful Citation
• 2. 1Password increases engineering productivity 21% with Codex
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Measuring LLM Sycophancy under Sustained Multi-Turn Pressure
• 5. A Data-Driven Framework for Identifying and Prioritizing RPA Opportunities in Healthcare Processes
• 6. How GPT-5.6 Sol helps run quantum computing experiments

Lumen Research Digest — 2026-09-08

2026-09-08 · 5 items

• 1. WorldSculpt: Generating Compositional Worlds from Grounded Videos
• 2. Legora reviewed 41 documents in minutes with GPT-6 Astra
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Distill Globally, Adapt Locally: Reasoning Distillation and Product-Type Test-Time Training for Scalable Trade-Up Recommendation
• 5. CONTINUITY: Security-Context Contracts for Composable LLM Agent Controls

Lumen Research Digest — 2026-09-07

2026-09-07 · 5 items

• 1. RoboSPA: Can VLA Models Go Beyond Simple Scenes and Short-Horizon Tasks?
• 2. Research acceleration: The view inside OpenAI
• 3. Broadening access to Skala creates a faster path to predictive DFT
• 4. Towards Neuro-Symbolic Procedural Reasoning for Long-Horizon Vision-Language-Action Manipulation
• 5. Compact Neural Appearance Models for Efficient Gaussian Splatting

Lumen Research Digest — 2026-09-06

2026-09-06 · 5 items

• 1. PatchBench: Evaluating AI Agents for Vulnerability Patching
• 2. Daybreak for Frontline Defenders: $1B to protect essential services
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. TokenMatch: 3D Mesh Correspondence Transformer with Curvature-Guided Tokenisation
• 5. Zero-Shot Novel Depth Synthesis Using 3D Foundation Models Scene Representations

Lumen Research Digest — 2026-09-05

2026-09-05 · 5 items

• 1. Formation Matrix and Energy-based Control of Multi-Agent Systems
• 2. GPT-6 Astra: A new generation of intelligence
• 3. Orchard: An open framework for scalable agentic AI
• 4. Scal3R: Learning Efficient Multi-Relative Pose Query for Scalable Online 3D Reconstruction
• 5. Puffin-World: Scaling a Unified Multimodal Model with Native 3D World States

Lumen Research Digest — 2026-09-04

2026-09-04 · 6 items

• 1. Principia: Relational Physics Tests for Video Models
• 2. Safety overview: GPT-6 Astra
• 3. EvoLib: Turning experience into evolving knowledge
• 4. SENTINEL-RL: Offloading Topological Reasoning from LLM Agents in the Security Operations Center
• 5. Beyond Retrieval: Progressive Latent Memory Evolution for Streaming Video Understanding
• 6. Playco cut manual fixes 50% prototyping games with GPT-6 Astra

Lumen Research Digest — 2026-09-03

2026-09-03 · 6 items

• 1. RoGe: Novel View Synthesis via End-to-End Implicit Reconstruction and Generation
• 2. How law firm Gilbert + Tobin governs and scales AI with OpenAI
• 3. Verifying Rust cryptography in SymCrypt, from standards to code
• 4. Towards Trustworthy Autonomous Robots: An Explainable AI-Based Decision Framework
• 5. Large Language Models (LLMs) for Telecom Root Cause Analysis (RCA): A Structured Reasoning Framework for Evidence-Grounded Diagnosis
• 6. ATV Big Air Tour turned 3 days of work into 3 hours with ChatGPT

Lumen Research Digest — 2026-09-02

2026-09-02 · 6 items

• 1. CordisBench: Can Language Models Reason About Component Lifecycles in Dynamic Agent Harnesses?
• 2. How AI-native companies turn workflows into operating capability
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. DualDiff3D: Dual Structure-Appearance Diffusion Priors for Reliability-Enhanced 3D Gaussian Splatting
• 5. TempCloze: Can Video-LLMs Identify the Missing Middle?
• 6. Path to Astra: critical capabilities and frontier safeguards

Lumen Research Digest — 2026-09-01

2026-09-01 · 5 items

• 1. Scaling Large Reasoning Models beyond Human Supervision: A Path toward Superintelligence
• 2. Polimill builds Japan's next-generation public AI infrastructure
• 3. GigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation models
• 4. BRF-GS: Hyperspectral Bidirectional Reflectance Factor Modeling and Image Generation Based on 3D Gaussian Splatting
• 5. Token-Efficient Data Reasoning Agents via Adaptive Structuring of Unstructured Data

Lumen Research Digest — 2026-08-31

2026-08-31 · 5 items

• 1. ChainSplat: A Physics-Inspired Screw-Theoretic Model for Learning Deformable Linear Object Dynamics from Multi-View RGB Videos
• 2. Expanding OpenAI’s presence in Brazil
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. AcrossVAM1.0: Particle World Modeling for Text-Assisted Robot Video Prediction
• 5. LLM-Based Agents for Software and Systems Security: Approaches, Applications, and Assessment

Lumen Research Digest — 2026-08-30

2026-08-30 · 5 items

• 1. Comparative Evaluation of 3D Reconstruction Methods for Immersive Visualization of Laboratory Objects
• 2. Better answers, broader thinking: What students gain from ChatGPT and critical-thinking training
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. CLAP: Cross-Embodiment Video World Models are Zero-Shot Physical Simulators
• 5. Reconstructing Humans and Objects in Interaction using Large Reconstruction Models

Lumen Research Digest — 2026-08-29

2026-08-29 · 5 items

• 1. UrbanGround: From Local Perception to Spatial Agency in a Real-Scale City
• 2. Learning never stops: How AI makes learning continuous
• 3. MindTopo reveals VLMs’ spatial reasoning abilities
• 4. Embodied Scene Rearrangement Planning
• 5. R2M-Bench: Evaluating Revisit Memory via Relative Consistency in Interactive Video World Models

Lumen Research Digest — 2026-08-28

2026-08-28 · 4 items

• 1. How loveholidays is making everyone a builder with Codex
• 2. Introducing CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement
• 3. The full stack behind abundant intelligence
• 4. Aurora 1.5: Extending open foundation models for weather and Earth-system applications

Lumen Research Digest — 2026-08-27

2026-08-27 · 6 items

• 1. VBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoning
• 2. Bringing ChatGPT for Teachers to more U.S. school districts
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. MyoMechanix: Biomechanically-Grounded Compositional Skilled Activity Understanding and Coaching
• 5. Answer Is Cheap, Show Me the Evidence! Augmenting Automated Vulnerability Assessment with Evidence
• 6. The Hugging Face incident and the road ahead

Lumen Research Digest — 2026-08-26

2026-08-26 · 6 items

• 1. BrowserForge: Scaling Web Episode via Parallel Browser Sandboxes
• 2. Jalapeño’s first results show industry-leading speed and efficiency in AI inference
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. FedV-KGQA: Multi-Hop Question Answering over Vertically Partitioned Knowledge Graphs
• 5. From Seeing to Acting: Smart Glasses as First-Person Intelligence Platforms
• 6. Introducing the Admin plugin for ChatGPT Work and Codex

Lumen Research Digest — 2026-08-25

2026-08-25 · 5 items

• 1. FixAnything: 3D-Consistent Rendering Refinement via Video Generative Priors
• 2. Advancing price-performance for developers with GPT‑5.6 in Kiro
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Act with Intent: Distilling Behavior Intent for Vision-Language-Action Models
• 5. SRPO: Self-Reflective Policy Optimization for Long-Horizon Reasoning

Lumen Research Digest — 2026-08-24

2026-08-24 · 5 items

• 1. ViTacPhys: Physical Property-Aware Grasping from Human Visual-Tactile Demonstrations
• 2. Introducing AI Futures
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. AI with Authority, from Application to Silicon
• 5. Beyond Fault Localization: A Trajectory-Level Study of LLM Agents for Microservice Root Cause Analysis

Lumen Research Digest — 2026-08-23

2026-08-23 · 5 items

• 1. Pandora's AI Model Routing Box: Efficient Allocation with Costly Value Estimation
• 2. ChatGPT Ads expands across Europe
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Rule-Compliant Visual Spatial Planning for Multimodal Large Language Models
• 5. The Third Restructuring of Software Form: From the Three-Tier Architecture to Storage, Models, and Agents

Lumen Research Digest — 2026-08-22

2026-08-22 · 5 items

• 1. AI4AI-Bench: Benchmarking LLM Agents in Algorithmic Design for Recursive Self-Improvement
• 2. Offering Zero Data Retention for frontier models
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. WithEveryone: Unified Planning and Identity Grounding for Group Image Generation
• 5. MidTool: Mid-training Data Synthesis for Agentic Tool Use

Lumen Research Digest — 2026-08-21

2026-08-21 · 6 items

• 1. DreamHand: Repurposing Video Diffusion Models for Occlusion-Robust Egocentric 3D Hand Motion Recovery
• 2. Broadening access to Skala creates a faster path to predictive DFT
• 3. Stampli cuts launch hours by 68% using ChatGPT Work
• 4. 4DAnyone: Create Anyone in 4D from a Casual Monocular Video
• 5. Video2DoorTraversal: Push Door Traversal via Simulated Door Twins
• 6. Replit expands access to software creation with GPT-5.6 Luna

Lumen Research Digest — 2026-08-20

2026-08-20 · 6 items

• 1. LT-Mem: Volatility-Aware Spatio-Temporal Memory for Lifelong Scene Understanding
• 2. Pacing model development in an era of cyber-critical capabilities
• 3. Orchard: An open framework for scalable agentic AI
• 4. SPADE: Self-Play in Adaptive Synthetic Executable Environments
• 5. What is Missing from AI Post-Training AI: An Empirical Analysis
• 6. Partnering with CodeAI to prepare the first AI generation

Lumen Research Digest — 2026-08-19

2026-08-19 · 6 items

• 1. GS-Voxel: Fitting-Free Structured Latents for Large-Scale 3DGS Generation
• 2. Asana cleared 5 years of engineering work in 2 weeks with Codex
• 3. Flint: A visualization language for the AI era
• 4. Deep Academic Survey: Stateful Agentic Closed-Loop Paradigm for Academic Survey Automation
• 5. Memory Tree Guided Key Frame Querying for Efficient 3D Question Answering
• 6. Strengthening democratic oversight in national security

Lumen Research Digest — 2026-08-18

2026-08-18 · 5 items

• 1. When State Becomes an Attack Surface: State-Semantic Injection in LLM-Driven Embodied Agents
• 2. The Defender’s Window
• 3. Verifying Rust cryptography in SymCrypt, from standards to code
• 4. Security of Foundation-Model-Powered Embodied Agents: Attack Surfaces, Attacks, Defenses, and Evaluation
• 5. FlexWorm: Primitive-augmented Hybrid Contact-motion Planning for Suction-based Multi-segment Deformable Robots

Lumen Research Digest — 2026-08-17

2026-08-17 · 5 items

• 1. Twin: Playing an Unknown Game with a Test-Time Digital Twin
• 2. OpenAI’s letter to Governor Abbott on responsible AI infrastructure in Texas
• 3. EvoLib: Turning experience into evolving knowledge
• 4. You Only Pass Once: Answering and Abstaining Together in a Single Forward Pass of a Frozen Language Model
• 5. Marionette: Predicting World States, Rendering Geometry, Painting Appearance

Lumen Research Digest — 2026-08-16

2026-08-16 · 5 items

• 1. AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design
• 2. OpenAI appoints Dali Rajic as Chief Revenue Officer
• 3. Echoverse: Deep, evolving environments for computer-use agents
• 4. Vero: Can AI Agents Build Formally Verified Software Repositories?
• 5. DreamX-Phi 1.0: Action-Conditioned Video World Model for Robotic Manipulation

Lumen Research Digest — 2026-08-15

2026-08-15 · 5 items

• 1. PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives
• 2. Testing ads in ChatGPT
• 3. Flint: A visualization language for the AI era
• 4. AlayaWorld: Interactive Long-Horizon World Modeling - Full Technical Report (v1.1)
• 5. GS$^{2}$CI: Robust Gaussian Splatting For Snapshot Compressive Imaging via Large Vision Model Priors

Lumen Research Digest — 2026-08-14

2026-08-14 · 6 items

• 1. Intern-S2-Preview: Scientific Agentic Foundation Model
• 2. The builder’s guide to GPT‑5.6
• 3. Flint: A visualization language for the AI era
• 4. OmniScientist: An Omni-Modal Omni-Discipline AI Scientist
• 5. TraVEL: Trajectory-Guided Video Embedding Learning for Driving-Video Retrieval
• 6. Previewing Ultrafast mode: GPT-5.6 Sol at up to 14X the speed

Lumen Research Digest — 2026-08-13

2026-08-13 · 6 items

• 1. AVA-Encoder: Towards Agent-Native Video Representation Learning
• 2. From assistance to execution: How enterprises put AI to work
• 3. MindTopo reveals VLMs’ spatial reasoning abilities
• 4. StateFlow: Building, Evolving, and Accessing 3D World States for Previsualization
• 5. Beyond Trial-and-Error: Agentic Optimization for Image-to-Video Adherence
• 6. How RingCentral builds AI-native work from engineering to ops

Lumen Research Digest — 2026-08-12

2026-08-12 · 4 items

• 1. Introducing CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement
• 2. Daybreak models are now available on AWS
• 3. Premium seats are coming to ChatGPT Business
• 4. Aurora 1.5: Extending open foundation models for weather and Earth-system applications

Lumen Research Digest — 2026-08-11

2026-08-11 · 6 items

• 1. Defining Decentralization: An Ontological Perspective
• 2. Expanding Daybreak as the Cyber Defense Window Narrows
• 3. SkillOpt: Agent skills as trainable parameters
• 4. Beyond Hazard Resemblance: Contrastive Event Adjudication for Training-Free Video Anomaly Detection
• 5. Learning How the World Evolves: Extrapolative Video World Models via Latent Dynamics Reasoning
• 6. Putting frontier cyber models in more trusted hands

Lumen Research Digest — 2026-08-10

2026-08-10 · 5 items

• 1. Toward a Causal Data Management Ecosystem for Decision Making and Agentic AI
• 2. From asking to doing: How the world is putting ChatGPT to work
• 3. Flint: A visualization language for the AI era
• 4. I Seek You in Videos: Identity-Conditioned Queries for Person-Centric Video Reasoning
• 5. SimWAM: A Simple World Action Model for End-to-End Autonomous Driving

Lumen Research Digest — 2026-08-09

2026-08-09 · 5 items

• 1. RP-OPSD: Reasoning-Pivot-Guided On-Policy Self-Distillation for Multilingual Reasoning Transfer
• 2. Working with the American Psychological Association on youth mental health and AI
• 3. Flint: A visualization language for the AI era
• 4. Depth-Guided Video Object Counting in Crowded Scenes
• 5. A Master-Salve Robot Manipulator for Needle-Based Teleoperation in MRI Chamber

Lumen Research Digest — 2026-08-08

2026-08-08 · 6 items

• 1. Learning When to Trust via Selective Context Preference Optimization
• 2. Responding to the next frontier of critical cyber capabilities
• 3. Flint: A visualization language for the AI era
• 4. Tracing the Heart: An Evidence-Linked Pipeline for Heart-Failure Feature Engineering
• 5. TRAJDEBUG: Tracing Error Lifecycle to Identify Critical Failures in Long-Horizon Agent Trajectories
• 6. How HSP GRUPPE builds AI capabilities for tax advisory

Lumen Research Digest — 2026-08-07

2026-08-07 · 5 items

• 1. GeniWorld: A Generalizable Interactive World Model for Robotic Manipulation via Visual Actions
• 2. Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users
• 3. Flint: A visualization language for the AI era
• 4. The Low Frequency Trap: Video Language Models Fail at Simple Event Bookkeeping
• 5. MASS: Multiplayer World Models with Authoritative Shared State

Lumen Research Digest — 2026-08-06

2026-08-06 · 5 items

• 1. SmartMage: Dynamic Modality Orchestration for 3D Scene Understanding
• 2. Apple is getting this wrong
• 3. Flint: A visualization language for the AI era
• 4. AI-based single-shot structured-light depth reconstruction for real-time laparoscopic surgical guidance
• 5. Objects as Audio-Visual Modal Sound Fields

Lumen Research Digest — 2026-08-05

2026-08-05 · 6 items

• 1. When and Where to Look: Adaptive Visual Evidence Scheduling for Efficient Long Video Understanding
• 2. Third-party cyber evaluations involving OpenAI models
• 3. Flint: A visualization language for the AI era
• 4. UniWorld-Design: From Pixel Generation to Layer-Native Design
• 5. Video-DeepResearch: Towards the Next-Generation Multimodal Deepresearch Agent
• 6. New ways to learn and teach with ChatGPT Work and Codex

Lumen Research Digest — 2026-08-04

2026-08-04 · 5 items

• 1. Antares: Foundation Models for Agentic Vulnerability Localization
• 2. Orchard: An open framework for scalable agentic AI
• 3. Circles powers telco personalization with OpenAI technology
• 4. Ego2Robot: Scalable Robot Data Synthesis from Egocentric Human Data
• 5. A Taxonomy of Cognitive Capability Gaps in Generative and Agentic AI

Lumen Research Digest — 2026-08-03

2026-08-03 · 5 items

• 1. CodeShrink: Adaptive Visual Compression for Efficient Multimodal Code Understanding
• 2. Ten advances in mathematics and theoretical computer science
• 3. Memora: A Harmonic Memory Representation Balancing Abstraction and Specificity
• 4. FibVLA: An Efficient Temporal Vision-Language-Action Model with Fibonacci Sampling
• 5. OASIS: Occlusion-aware Single-image Hand Avatar Reconstruction via 3D Gaussian Splatting

Lumen Research Digest — 2026-08-02

2026-08-02 · 5 items

• 1. Change2Task: From Repository Changes to Executable Coding Agent Tasks and Environments
• 2. Advancing the price-performance frontier with GPT-5.6
• 3. Flint: A visualization language for the AI era
• 4. ReToken: One Token to Improve Vision-Language Models for Visual Retrieval
• 5. DualG-MRAG: Decoupling Macro-Reasoning and Micro-Matching for Multimodal Retrieval-Augmented Generation

Lumen Research Digest — 2026-08-01

2026-08-01 · 5 items

• 1. OSReward: Instituting Standardized Evaluation for Cross-Platform Computer-Use Reward Models
• 2. Advancing responsible AI across Europe
• 3. Verifying Rust cryptography in SymCrypt, from standards to code
• 4. X-NavDP: Generalizing Navigation Diffusion Policy to Novel Behavior and Embodiments with Group Q-score Reweighted Matching
• 5. ACE-Data-0: Human-Centric Ambient Capture as Embodied Data Engine

Lumen Research Digest — 2026-07-31

2026-07-31 · 6 items

• 1. FA-RDP: A Frequency-Adaptive Reactive Diffusion Policy for Contact-Rich Manipulation
• 2. Echoverse: Deep, evolving environments for computer-use agents
• 3. How avatarin built a 24/7 retail agent with GPT-Realtime
• 4. ORCA-bench: How Ready Are Language Model Agents for Oncall?
• 5. PhiZero: A World Model Built Around Physical Language
• 6. EvoLib: Turning experience into evolving knowledge

Lumen Research Digest — 2026-07-30

2026-07-30 · 6 items

• 1. Explainable and Resource-Efficient Spatial Reasoning in Multimodal LLMs for Decision-Critical Applications
• 2. How GPT-5.6 fuses frontier intelligence with frontier efficiency
• 3. Understanding the brain with AI-driven explanations and experiments
• 4. AgentMap: Joint Equivalence and Subsumption Discovery for Ontology Matching
• 5. MemSecBench: Tracking Agent Memory Poisoning from Persistence to Consequence and Repair
• 6. How enabling two settings tripled our scores on the ARC-AGI-3 benchmark

Lumen Research Digest — 2026-07-29

2026-07-29 · 5 items

• 1. Wonder: Video World Model Done Better
• 2. Scientific computing in the age of agentic AI
• 3. Talos: Scaling rare disease diagnosis with automated, iterative genomic reanalysis
• 4. Knowledge-Guided Multimodal Reasoning over Interacting Streams for Video-Level Ambivalence and Hesitancy Recognition
• 5. Schrödinger's Cat: Probabilistic Representation and Prediction of Potential Scene Kinematics

Lumen Research Digest — 2026-07-28

2026-07-28 · 5 items

• 1. Data Pyramid for Embodied Manipulation
• 2. Launching Health in ChatGPT
• 3. Flint: A visualization language for the AI era
• 4. ClinFusion: A Vision-Centric Multimodal LLM System for Holistic Medical Understanding
• 5. ERUnderstand: Evaluating Vision-Language Models on Structured ER Diagrams

Lumen Research Digest — 2026-07-27

2026-07-27 · 4 items

• 1. Introducing the ChatGPT for small business program
• 2. Aurora 1.5: Extending open foundation models for weather and Earth-system applications
• 3. Flint: A visualization language for the AI era
• 4. Introducing OpenAI Presence

Lumen Research Digest — 2026-07-26

2026-07-26 · 5 items

• 1. Unified Video Dense Prediction from Disjoint Data
• 2. Advancing the next era of national science
• 3. Flint: A visualization language for the AI era
• 4. MedGame: Storytelling Gamification Empowered by Large Language Models for Medical Education
• 5. Agentic Context Management: Solving Agent Memory and Cost by Treating Them as Lifecycle and Architecture Problems

Lumen Research Digest — 2026-07-25

2026-07-25 · 5 items

• 1. 3D-Aware VLMs with Implicit and Explicit Geometries
• 2. How news organizations are using AI to advance their vital missions
• 3. SkillOpt: Agent skills as trainable parameters
• 4. Beyond Episodic Evaluation: Memory Architectural Bottlenecks in Sequential Embodied Question Answering
• 5. Streaming Multi-Agent Autoregressive Diffusion Model with World State Registers

Lumen Research Digest — 2026-07-24

2026-07-24 · 5 items

• 1. OpenForgeRL: Train Harness-native Agents in Any Environment
• 2. Building AI infrastructure with the Effingham County community
• 3. Flint: A visualization language for the AI era
• 4. GS-Agent: Creating 4D Physical Worlds With Generative Simulation
• 5. Euclid-MCP: A Model Context Protocol Server for Deterministic Logical Reasoning via Prolog

Lumen Research Digest — 2026-07-23

2026-07-23 · 6 items

• 1. Multimodal Large Language Models for Remote Sensing Image Understanding: Domain-Specific or General-Purpose?
• 2. Introducing OpenAI Presence
• 3. Flint: A visualization language for the AI era
• 4. ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion
• 5. MR-Compare: A Mixed-Reality Framework for Spatially Grounded Visual Comparison of 3D Gaussian Splatting and Mesh Reconstructions with the Physical Environment
• 6. NTT DATA Group cuts incident analysis to 30 minutes with Codex

Lumen Research Digest — 2026-07-22

2026-07-22 · 5 items

• 1. They'll Verify. They Just Won't Act. How Authority Framing and Laundered Code Turn a Trusted Agentic CI/CD Pipeline Into an Attack Surface
• 2. OpenAI and Hugging Face partner to address security incident during model evaluation
• 3. Flint: A visualization language for the AI era
• 4. MeetingToM: Evaluating Multimodal LLMs on Theory-of-Mind Reasoning in Multi-Party Meetings
• 5. Agents in the Wild: Where Research Meets Deployment

Lumen Research Digest — 2026-07-21

2026-07-21 · 5 items

• 1. FlashRT: Agent Harness for Guiding Agents to Deploy Real-Time Multimodal Applications
• 2. Safety and alignment in an era of long-horizon models
• 3. Flint: A visualization language for the AI era
• 4. Robust Multimodal Dynamic Object Segmentation
• 5. O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning

Lumen Research Digest — 2026-07-20

2026-07-20 · 5 items

• 1. Knowing the Self, Understanding the World: A Dual-Cognition Benchmark for UAV Spatio-temporal Reasoning with MLLMs
• 2. Why teens deserve access to safe AI
• 3. Ire identifies another LOTUSLITE specimen
• 4. The Honest Quorum Problem: Epistemic Byzantine Fault Tolerance for Agentic Infrastructure
• 5. Vision-Language-Motion Maps: An Open-Vocabulary, Uncertainty-Aware, Queryable Motion Attribute for 3D Scene Maps

Lumen Research Digest — 2026-07-19

2026-07-19 · 5 items

• 1. RoboTTT: Context Scaling for Robot Policies
• 2. GPT-Red: Unlocking Self-Improvement for Robustness
• 3. Flint: A visualization language for the AI era
• 4. SearchOS-V1: Towards Robust Open-Domain Information-Seeking Agent Collaboration
• 5. MAGiSt3R: Multi-Agent Feed-forward 3D Reconstruction from Monocular RGB Videos

Lumen Research Digest — 2026-07-18

2026-07-18 · 6 items

• 1. Bridge Evidence: Static Retrieval Utility Does Not Predict Causal Utility in Multi-Step Agentic Search
• 2. How Cars24 scales conversations and builds faster with OpenAI
• 3. Flint: A visualization language for the AI era
• 4. Beyond the Leaderboard: Design Lessons for Trustworthy Multimodal VQA
• 5. Setup Complete, Now You Are Compromised: Weaponizing Setup Instructions Against AI Coding Agents
• 6. A scorecard for the AI age

Lumen Research Digest — 2026-07-17

2026-07-17 · 4 items

• 1. Beyond Success Rate: Cost-Aware Evaluation of Offensive and Defensive Security Agents
• 2. Memora: A Harmonic Memory Representation Balancing Abstraction and Specificity
• 3. MM-IssueLoc: A Controlled Benchmark for Evaluating Visual Evidence in Multimodal Repository-Level Issue Localization
• 4. Hierarchical Denoising For Multi-Step Visual Reasoning

Lumen Research Digest — 2026-07-16

2026-07-16 · 5 items

• 1. ProfMalPlus: Agent-Coordinated Detection of Malicious NPM Packages via Static-Dynamic Analysis Synergy
• 2. The US is advancing AI safety through state and federal action
• 3. Flint: A visualization language for the AI era
• 4. Industrial Dexterity Benchmark: A Hardware-Software Benchmarking Platform for Industrial Dexterous Manipulation
• 5. VisualRepair: Dynamic Tool Calling and Region Focusing for Visual Software Issue Repair

Lumen Research Digest — 2026-07-15

2026-07-15 · 6 items

• 1. MAMMOTH: A Multi-Modal End-to-End Policy for Off-Road Mobility Robust to Missing Modality
• 2. How to manage AI investments in the agentic era
• 3. Flint: A visualization language for the AI era
• 4. Do AI Agents Know When a Task Is Simple? Toward Complexity-Aware Reasoning and Execution
• 5. TerraZero: Procedural Driving Simulation for Zero-Demonstration Self-Play at Scale
• 6. How data science teams use ChatGPT Work

Lumen Research Digest — 2026-07-14

2026-07-14 · 5 items

• 1. Beyond the Single Camera: Agentic Multi-View Reasoning in Sports Video Understanding
• 2. Verifying Rust cryptography in SymCrypt, from standards to code
• 3. Getting started with ChatGPT
• 4. Casting Everything to Online API Services? A Survey of Integrating Localized Speech Recognition Models in Robotic Systems
• 5. When Local Monitors Miss Compositional Harm: Diagnosing Distributed Backdoors in Multi-Agent Systems

Lumen Research Digest — 2026-07-13

2026-07-13 · 5 items

• 1. Task-Specific Multimodal Question Answering Agents via Confidence Calibration and Incremental Reasoning for QANTA 2026
• 2. How Deutsche Telekom is rewiring telecommunications with AI
• 3. Understanding the brain with AI-driven explanations and experiments
• 4. 4DR360: State Reasoning for Joint 3D Detection and Occupancy Prediction in 4D Radar-Camera Full-Scene Perception
• 5. VEXAIoT: Autonomous IoT Vulnerability EXploitation using AI Agents

Lumen Research Digest — 2026-07-12

2026-07-12 · 5 items

• 1. UniClawBench: A Universal Benchmark for Proactive Agents on Real-World Tasks
• 2. GPT-5.6: Frontier intelligence that scales with your ambition
• 3. Talos: Scaling rare disease diagnosis with automated, iterative genomic reanalysis
• 4. Wat3R: Underwater 3D Geometry Learning without Annotations
• 5. ProjAgent: Procedural Similarity Retrieval for Repository-Level Code Generation

Lumen Research Digest — 2026-07-11

2026-07-11 · 5 items

• 1. Geometry and Gradient-based Partitioning for Panoramic Outdoor Reconstruction
• 2. Helping K–12 educators build practical AI skills
• 3. Flint: A visualization language for the AI era
• 4. AUTOPILOT VQA: Benchmarking Vision-Language Models for Incident-Centric Dashcam Understanding
• 5. ARDY: Autoregressive Diffusion with Hybrid Representation for Interactive Human Motion Generation

Lumen Research Digest — 2026-07-10

2026-07-10 · 4 items

• 1. ChatGPT is now a partner for your most ambitious work
• 2. Aurora 1.5: Extending open foundation models for weather and Earth-system applications
• 3. GPT-5.6 is now the preferred model in Microsoft 365 Copilot
• 4. Flint: A visualization language for the AI era

Lumen Research Digest — 2026-07-09

2026-07-09 · 6 items

• 1. CARLA-GS: Decoupling Representation, Reasoning, and Physics Simulation for Autonomous Driving Corner-Case Synthesis
• 2. Flint: A visualization language for the AI era
• 3. Our approach to government and national security partnerships
• 4. Agon: Competitive Cross-Model RL with Implicit Rival Grading of Reasoning
• 5. Infinite Worlds with Versatile Interactions
• 6. Separating signal from noise in coding evaluations

Lumen Research Digest — 2026-07-08

2026-07-08 · 5 items

• 1. RynnWorld-4D: 4D Embodied World Models for Robotic Manipulation
• 2. Australian Payments Plus moves faster with ChatGPT and Codex
• 3. SkillOpt: Agent skills as trainable parameters
• 4. CAIRN: Cross-Room 3D Scene Understanding with Topology-Aware Large Multimodal Models
• 5. RynnWorld-Teleop: An Action-Conditioned World Model for Digital Teleoperation

Lumen Research Digest — 2026-07-07

2026-07-07 · 5 items

• 1. LLM-as-a-Verifier: A General-Purpose Verification Framework
• 2. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models
• 3. How agents are transforming work
• 4. InFlux++: Real and Synthetic Data for Estimating Dynamic Camera Intrinsics
• 5. GaP: A Graph-as-Policy Multi-Agent Self-Learning Harness For Variational Automation Tasks

Lumen Research Digest — 2026-07-06

2026-07-06 · 5 items

• 1. Controllable Sim Agents with Behavior Latents
• 2. SkillOpt: Agent skills as trainable parameters
• 3. How agents are transforming work
• 4. GeoMix: Descriptor-Free Visual Localization via Global Context and Multi-Detector Training
• 5. ReContext: Recursive Evidence Replay as LLM Harness for Long-Context Reasoning

Lumen Research Digest — 2026-07-05

2026-07-05 · 5 items

• 1. Reasoning effort, not tool access, buys first-try reliability in agentic code generation: an observational study
• 2. SkillOpt: Agent skills as trainable parameters
• 3. How agents are transforming work
• 4. Steerability via constraints: a substrate for scalable oversight of coding agents
• 5. Distributed Attacks in Persistent-State AI Control

Lumen Research Digest — 2026-07-04

2026-07-04 · 5 items

• 1. Understanding Agent-Based Patching of Compiler Missed Optimizations
• 2. SkillOpt: Agent skills as trainable parameters
• 3. How agents are transforming work
• 4. Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots
• 5. TestEvo-Bench: An Executable and Live Benchmark for Test and Code Co-Evolution

Lumen Research Digest — 2026-07-03

2026-07-03 · 5 items

• 1. Reasoning LLM Improves Speaker Recognition in Long-form TV Dramas
• 2. HP Inc. launches Frontier strategic partnership with OpenAI
• 3. Ire identifies another LOTUSLITE specimen
• 4. Alignment Is All You Need For X-to-4D Generation
• 5. WorldDirector: Building Controllable World Simulators with Persistent Dynamic Memory

Lumen Research Digest — 2026-07-02

2026-07-02 · 5 items

• 1. Structured 4D Latent Predictive Model for Robot Planning
• 2. Core dump epidemiology: fixing an 18-year-old bug
• 3. Extending Human Intelligence Through AI
• 4. World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video
• 5. RepoRescue: An Empirical Study of LLM Agents on Whole-Repository Compatibility Rescue

Lumen Research Digest — 2026-07-01

2026-07-01 · 6 items

• 1. PointSplat: Compact Gaussian Splatting via Human-Centric Prediction
• 2. SkillOpt: Agent skills as trainable parameters
• 3. Introducing GeneBench-Pro
• 4. DVG-WM: Disentangled Video Generation Enables Efficient Embodied World Model for Robotic Manipulation
• 5. ERA: Entropy-Guided Visual Token Pruning with Rectified Attention for Efficient MLLMs
• 6. How ChatGPT adoption has expanded

Lumen Research Digest — 2026-06-30

2026-06-30 · 5 items

• 1. VLK: Learning Humanoid Loco-Manipulation from Synthetic Interactions in Reconstructed Scenes
• 2. Memora: A Harmonic Memory Representation Balancing Abstraction and Specificity
• 3. Mapping Europe’s AI Workforce Opportunity
• 4. Open-Vocabulary and Referring Segmentation for 3D Gaussians Using 2D Detectors
• 5. UnfoldArt: Zero-Shot Recovery of Full Articulated 3D Objects from Text or Image

Lumen Research Digest — 2026-06-29

2026-06-29 · 5 items

• 1. StructSplat: Generalizable 3D Gaussian Splatting from Uncalibrated Sparse Views
• 2. How Omio is building the future of conversational travel
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. Agent-Native Immune System: Architecture, Taxonomy, and Engineering
• 5. HAT-4D: Lifting Monocular Video for 4D Multi-Object Interactions via Human-Agent Collaboration

Lumen Research Digest — 2026-06-28

2026-06-28 · 5 items

• 1. Empowering GUI Agents via Autonomous Experience Exploration and Hindsight Experience Utilization for Task Planning
• 2. Helping build shared standards for advanced AI
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. Bridging Talk and Thought: Understanding Dialogue Dynamics Across Collaborative Problem-Solving Contexts
• 5. Autoregressive Boltzmann Generators

Lumen Research Digest — 2026-06-27

2026-06-27 · 5 items

• 1. Advancing Omnimodal Embodied Agents from Isolated Skills to Everyday Physical Autonomy
• 2. Previewing GPT-5.6 Sol: a next-generation model
• 3. Vega: Zero-knowledge proofs for digital identity in the age of AI
• 4. NOVA: A Verification-Aware Agent Harness for Architecture Evolution in Industrial Recommender Systems
• 5. Paying More Attention to Visual Tokens in Self-Evolving Large Multimodal Models

Lumen Research Digest — 2026-06-26

2026-06-26 · 5 items

• 1. PhysiFormer: Learning to Simulate Mechanics in World Space
• 2. How agents are transforming work
• 3. Understanding the brain with AI-driven explanations and experiments
• 4. OctoSense: Self-Supervised Learning for Multimodal Robot Perception
• 5. E-TTS: A New Embodied Test-Time Scaling Framework for Robotic Manipulation

Lumen Research Digest — 2026-06-25

2026-06-25 · 5 items

• 1. RoboAtlas: Contextual Active SLAM
• 2. OpenAI and Broadcom unveil LLM-optimized inference chip
• 3. Talos: Scaling rare disease diagnosis with automated, iterative genomic reanalysis
• 4. TriViewBench: Controlled Complexity Scaling for Multi-View Structural Reasoning in MLLMs
• 5. Same Evidence, Different Answer: Auditing Order Sensitivity in Multimodal Large Language Models

Lumen Research Digest — 2026-06-24

2026-06-24 · 6 items

• 1. Pocket-SLAM: Rendering-Area-Aware Pruning for Memory-Efficient 3DGS-SLAM
• 2. Patch the Planet: a Daybreak initiative to support open source maintainers
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. OrbitForge: Text-to-3D Scene Generation via Reconstruction-Anchored Video Synthesis
• 5. Accuracy and Satisfaction in Multi-Turn LLM Dialogues for NFR Assessment
• 6. How GPT-5 helped immunologist Derya Unutmaz solve a 3-year-old mystery

Lumen Research Digest — 2026-06-23

2026-06-23 · 6 items

• 1. Kamera: Unified Position-Invariant Multimodal KV Cache for Training-Free Reuse
• 2. Daybreak: Tools for securing every organization in the world
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. Lift4D: Harmonizing Single-View 3D Estimation for 4D Reconstruction In-the-Wild
• 5. AIR: Adaptive Interleaved Reasoning with Code in MLLMs
• 6. Codex-maxxing for long-running work

Lumen Research Digest — 2026-06-22

2026-06-22 · 5 items

• 1. Calibration Without Comprehension: Diagnosing the Limits of Fine-Tuning LLMs for Vulnerability Detection in Systems Software
• 2. Samsung Electronics brings ChatGPT and Codex to employees
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. Analyzing Defensive Misdirection Against Model-Guided Automated Attacks on Agentic AI Systems
• 5. Multi-LCB: Extending LiveCodeBench to Multiple Programming Languages

Lumen Research Digest — 2026-06-21

2026-06-21 · 5 items

• 1. Sovereign Execution Brokers: Enforcing Certificate-Bound Authority in Agentic Control Planes
• 2. GridSFM: A new, small foundation model for the electric grid
• 3. Introducing LifeSciBench
• 4. TimeProVe: Propose, then Verify for Efficient Long Video Temporal Reasoning in Activities of Daily Living
• 5. Efficient and Sound Probabilistic Verification for AI Agents

Lumen Research Digest — 2026-06-20

2026-06-20 · 5 items

• 1. Thinking in Boxes: 3D Editing in Real Images Made Easy
• 2. New usage analytics and updated spend controls for enterprises
• 3. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models
• 4. SARLO-80: Worldwide Slant SAR Language Optic Dataset 80cm
• 5. LedgerAgent: Structured State for Policy-Adherent Tool-Calling Agents

Lumen Research Digest — 2026-06-19

2026-06-19 · 6 items

• 1. Current World Models Lack a Persistent State Core
• 2. Improving health intelligence in ChatGPT
• 3. Further Notes on Our Recent Research on AI Delegation and Long-Horizon Reliability
• 4. DeepSWIP: Quotient-WMC Counterfactuals for Neural Probabilistic Logic Programs
• 5. Execution-State Capsules: Graph-Bound Execution-State Checkpoint and Restore for Low-Latency, Small-Batch, On-Device Physical-AI Serving
• 6. Using AI to help physicians diagnose rare genetic diseases affecting children

Lumen Research Digest — 2026-06-18

2026-06-18 · 6 items

• 1. OneCanvas: 3D Scene Understanding via Panoramic Reprojection
• 2. Introducing LifeSciBench
• 3. mimalloc: A new, high-performance, scalable memory allocator for the modern era
• 4. Native Active Perception as Reasoning for Omni-Modal Understanding
• 5. A Mixed-Reality Testbed for Autonomous Vehicles
• 6. A near-autonomous AI chemist improves a challenging reaction in medicinal chemistry

Lumen Research Digest — 2026-06-17

2026-06-17 · 5 items

• 1. EgoCS-400K: An Egocentric Gameplay Dataset for World Models
• 2. Predicting model behavior before release by simulating deployment
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. Seeing Is Not Screening: Multimodal Hidden Instruction Attacks on Agent Skill Scanners
• 5. Future Dynamic 3D Reconstruction: A 3D World Model with Disentangled Ego-Motion

Lumen Research Digest — 2026-06-16

2026-06-16 · 5 items

• 1. Geometric Action Model for Robot Policy Learning
• 2. How Preply combines AI and human tutors to personalize learning
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. R2RDreamer: 3D-aware Data Augmentation for Spatially-generalized 2D Manipulation Policies
• 5. Context-Aware RL for Agentic and Multimodal LLMs

Lumen Research Digest — 2026-06-15

2026-06-15 · 5 items

• 1. AgentSpec: Understanding Embodied Agent Scaffolds Through Controlled Composition
• 2. Introducing the OpenAI Partner Network
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. OmniVideo-100K: A Dataset for Audio-Visual Reasoning through Structured Scripts and Evidence Chains
• 5. RepFusion: Leveraging Multimodal Priors for Denoising in Representation Space

Lumen Research Digest — 2026-06-14

2026-06-14 · 5 items

• 1. World Tracing: Generative Pixel-Aligned Geometry Beyond the Visible
• 2. BBVA puts AI at the core of banking with OpenAI
• 3. Extending Human Intelligence Through AI
• 4. Beyond Runtime Enforcement: Shield Synthesis as Defensibility Analysis for Adversarial Networks
• 5. Surflo: Consistent 3D Surface Flow Model with Global State

Lumen Research Digest — 2026-06-13

2026-06-13 · 6 items

• 1. $\texttt{WEAVER}$, Better, Faster, Longer: An Effective World Model for Robotic Manipulation
• 2. New OpenAI Academy courses for the next era of work
• 3. Ire identifies another LOTUSLITE specimen
• 4. RepWAM: World Action Modeling with Representation Visual-Action Tokenizers
• 5. Agents-K1: Towards Agent-native Knowledge Orchestration
• 6. PRC-linked influence operations are targeting AI debates in the US

Lumen Research Digest — 2026-06-12

2026-06-12 · 6 items

• 1. InterleaveThinker: Reinforcing Agentic Interleaved Generation
• 2. OpenAI to acquire Ona
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. SpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning
• 5. Flex4DHuman: Flexible Multi-view Video Diffusion for 4D Human Reconstruction
• 6. Supporting Europe’s work in ensuring a trustworthy AI ecosystem

Lumen Research Digest — 2026-06-11

2026-06-11 · 6 items

• 1. DIRECT: When and Where Should You Allocate Test-Time Compute in Embodied Planners?
• 2. Access OpenAI models and Codex through your Oracle cloud commitment
• 3. Vega: Zero-knowledge proofs for digital identity in the age of AI
• 4. A Five-Plane Reference Architecture for Runtime Governance of Production AI Agents
• 5. Breaking Entropy Bounds: Accelerating RL Training via MTP with Rejection Sampling
• 6. How an astrophysicist uses Codex to help simulate black holes

Lumen Research Digest — 2026-06-10

2026-06-10 · 6 items

• 1. WorldOlympiad: Can Your World Model Survive a Triathlon?
• 2. How engineers at Nextdoor use Codex to build without limits
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. ABC-Bench: An Agentic Bio-Capabilities Benchmark for Biosecurity
• 5. P3D-Bench: Benchmarking MLLMs for Parametric 3D Generation and Structural Reasoning
• 6. What Codex unlocks for Notion

Lumen Research Digest — 2026-06-09

2026-06-09 · 5 items

• 1. Latent Spatial Memory for Video World Models
• 2. Built to benefit everyone: our plan
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. iMaC: Translating Actions into Motion and Contact Images for Embodied World Models
• 5. Observability for Delegated Execution in Agentic AI Systems

Lumen Research Digest — 2026-06-08

2026-06-08 · 5 items

• 1. MemDreamer: Decoupling Perception and Reasoning for Long Video Understanding via Hierarchical Graph Memory and Agentic Retrieval Mechanism
• 2. Dreaming: Better memory for a more helpful ChatGPT
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. UniSHARP: Universal Sharp Monocular View Synthesis
• 5. Watch, Remember, Reason: Human-View Video Understanding with MLLMs

Lumen Research Digest — 2026-06-07

2026-06-07 · 5 items

• 1. You Only Index Once: Cross-Layer Sparse Attention with Shared Routing
• 2. OpenAI public policy agenda
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. Robust Ensemble of Selectively Strengthened and Augmented Predictors
• 5. DNQ: Deep Nash Q-Network for Partially Observable n-Player Games

Lumen Research Digest — 2026-06-06

2026-06-06 · 5 items

• 1. PAR3D: A Unified 3D-MLLM with Part-Aware Representation for Scene Understanding
• 2. Travelers deploys AI-powered claims countrywide with OpenAI
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. Visual Commonsense Driven Knowledge Refinements for Scene Graph Generation
• 5. Thinking with Imagination: Agentic Visual Spatial Reasoning with World Simulators

Lumen Research Digest — 2026-06-05

2026-06-05 · 6 items

• 1. Benchmark Everything Everywhere All at Once
• 2. How Endava is redesigning software delivery around AI agents
• 3. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 4. StoryVideoQA: Scaling Deep Video Understanding with a Large-Scale, Multi-Genre and Auto-Generated Dataset
• 5. Will the Agent Recuse Itself? Measuring LLM-Agent Compliance with In-Band Access-Deny Signals
• 6. How Wasmer used Codex to build a Node.js runtime for the edge

Lumen Research Digest — 2026-06-04

2026-06-04 · 4 items

• 1. Introducing new capabilities to GPT-Rosalind
• 2. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 3. A blueprint for democratic governance of frontier AI
• 4. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models

Lumen Research Digest — 2026-06-03

2026-06-03 · 4 items

• 1. Codex is becoming a productivity tool for everyone
• 2. Further Notes on Our Recent Research on AI Delegation and Long-Horizon Reliability
• 3. Codex for every role, tool, and workflow
• 4. Data Formulator 0.7: AI-powered data analytics for enterprise data

Lumen Research Digest — 2026-06-02

2026-06-02 · 4 items

• 1. OpenAI frontier models and Codex are now available on AWS
• 2. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 3. Building the infrastructure for the Intelligence Age in Michigan
• 4. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models

Lumen Research Digest — 2026-06-01

2026-06-01 · 4 items

• 1. How Braintrust turns customer requests into code with Codex
• 2. mimalloc: A new, high-performance, scalable memory allocator for the modern era
• 3. GridSFM: A new, small foundation model for the electric grid
• 4. How Endava builds an agentic organization with Codex

Lumen Research Digest — 2026-05-31

2026-05-31 · 6 items

• 1. DynaFLIP: Rethinking Robotics Perception via Tri-Modal-Dynamics Guided Representation
• 2. Boston Children’s uses AI to unlock new diagnoses
• 3. Advancing AI for materials with MatterSim: experimental synthesis, faster simulation, and multi-task models
• 4. RoboWits: Unexpected Challenges for Robotic Creative Problem Solving
• 5. GMOS: Grounding Moving Object Segmentation in 3D Space and Time
• 6. Strengthening societal resilience with Rosalind Biodefense

Lumen Research Digest — 2026-05-30

2026-05-30 · 4 items

• 1. A shared playbook for trustworthy third party evaluations
• 2. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 3. Cisco and OpenAI redefine enterprise engineering with Codex
• 4. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models

Lumen Research Digest — 2026-05-29

2026-05-29 · 4 items

• 1. How Endava builds an agentic organization with Codex
• 2. Data Formulator 0.7: AI-powered data analytics for enterprise data
• 3. OpenAI’s Frontier Governance Framework
• 4. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models

Lumen Research Digest — 2026-05-28

2026-05-28 · 4 items

• 1. Building self-improving tax agents with Codex
• 2. Extending Human Intelligence Through AI
• 3. Warp’s big bet on building open source with GPT-5.5
• 4. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests

Lumen Research Digest — 2026-05-27

2026-05-27 · 4 items

• 1. OpenAI, Grupo Folha and Grupo UOL announce strategic content partnership
• 2. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 3. OpenAI named a Leader in enterprise coding agents by Gartner
• 4. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models

Lumen Research Digest — 2026-05-26

2026-05-26 · 5 items

• 1. TriSplat: Simulation-Ready Feed-Forward 3D Scene Reconstruction
• 2. AdventHealth advances whole-person care with OpenAI
• 3. Building realistic electric transmission grid dataset at scale: a pipeline from open dataset
• 4. Squeezing Capacity from Multimodal Large Language Models for Subject-driven Generation
• 5. AnyScene: Towards Highly Controllable Driving Scene Generation at Anywhere and Beyond

Lumen Research Digest — 2026-05-25

2026-05-25 · 5 items

• 1. ETCHR: Editing To Clarify and Harness Reasoning
• 2. Advancing content provenance for a safer, more transparent AI ecosystem
• 3. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 4. Agentic Proving for Program Verification
• 5. SkillOpt: Executive Strategy for Self-Evolving Agent Skills

Lumen Research Digest — 2026-05-24

2026-05-24 · 5 items

• 1. Sensor2Sensor: Cross-Embodiment Sensor Conversion for Autonomous Driving
• 2. An OpenAI model has disproved a central conjecture in discrete geometry
• 3. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 4. LCGuard: Latent Communication Guard for Safe KV Sharing in Multi-Agent Systems
• 5. MotiMotion: Motion-Controlled Video Generation with Visual Reasoning

Lumen Research Digest — 2026-05-23

2026-05-23 · 6 items

• 1. GesVLA: Gesture-Aware Vision-Language-Action Model Embedded Representations
• 2. OpenAI named a Leader in enterprise coding agents by Gartner
• 3. Vega: Zero-knowledge proofs for digital identity in the age of AI
• 4. Cambrian-P: Pose-Grounded Video Understanding
• 5. AwareVLN: Reasoning with Self-awareness for Vision-Language Navigation
• 6. How Virgin Atlantic ships faster with Codex

Lumen Research Digest — 2026-05-22

2026-05-22 · 4 items

• 1. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models
• 2. The next phase of OpenAI’s Education for Countries
• 3. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 4. OpenAI and Dell partner to bring Codex to hybrid and on-premise enterprise environments

Lumen Research Digest — 2026-05-21

2026-05-21 · 5 items

• 1. DeepWeb-Bench: A Deep Research Benchmark Demanding Massive Cross-Source Evidence and Long-Horizon Derivation
• 2. How Ramp engineers accelerate code review with Codex
• 3. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 4. You Only Need Minimal RLVR Training: Extrapolating LLMs via Rank-1 Trajectories
• 5. Quality and Security Signals in AI-Generated Python Refactoring Pull Requests

Lumen Research Digest — 2026-05-20

2026-05-20 · 5 items

• 1. ClinSeekAgent: Automating Multimodal Evidence Seeking for Agentic Clinical Reasoning
• 2. Introducing OpenAI for Singapore
• 3. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 4. What Do Evolutionary Coding Agents Evolve?
• 5. TideGS: Scalable Training of Over One Billion 3D Gaussian Splatting Primitives via Out-of-Core Optimization

Lumen Research Digest — 2026-05-19

2026-05-19 · 5 items

• 1. Code as Agent Harness
• 2. OpenAI and Dell partner to bring Codex to hybrid and on-premise enterprise environments
• 3. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 4. Advancing Narrative Long Video Generation via Training-Free Identity-Aware Memory
• 5. ESI-Bench: Towards Embodied Spatial Intelligence that Closes the Perception-Action Loop

Lumen Research Digest — 2026-05-18

2026-05-18 · 6 items

• 1. Argus: Evidence Assembly for Scalable Deep Research Agents
• 2. OpenAI and Malta partner to bring ChatGPT Plus to all citizens
• 3. Red-teaming a network of agents: Understanding what breaks when AI agents interact at scale
• 4. Confirming Correct, Missing the Rest: LLM Tutoring Agents Struggle Where Feedback Matters Most
• 5. IVGT: Implicit Visual Geometry Transformer for Neural Scene Representation
• 6. How data science teams use Codex

Lumen Research Digest — 2026-05-17

2026-05-17 · 4 items

• 1. A new personal finance experience in ChatGPT
• 2. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 3. How business operations teams use Codex
• 4. Microsoft at NSDI 2026: Advances in large-scale networked systems

Lumen Research Digest — 2026-05-16

2026-05-16 · 6 items

• 1. Pelican-Unified 1.0: A Unified Embodied Intelligence Model for Understanding, Reasoning, Imagination and Action
• 2. Databricks brings GPT-5.5 to enterprise agent workflows
• 3. Further Notes on Our Recent Research on AI Delegation and Long-Horizon Reliability
• 4. APWA: A Distributed Architecture for Parallelizable Agentic Workflows
• 5. Articraft: An Agentic System for Scalable Articulated 3D Asset Generation
• 6. Work with Codex from anywhere

Lumen Research Digest — 2026-05-15

2026-05-15 · 4 items

• 1. Sea's View on the Future of Agentic Software Development with Codex
• 2. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 3. How finance teams use Codex
• 4. Microsoft at NSDI 2026: Advances in large-scale networked systems

Lumen Research Digest — 2026-05-14

2026-05-14 · 7 items

• 1. EVA-Bench: A New End-to-end Framework for Evaluating Voice Agents
• 2. Building a safe, effective sandbox to enable Codex on Windows
• 3. mimalloc: A new, high-performance, scalable memory allocator for the modern era
• 4. Harnessing Agentic Evolution
• 5. Good Agentic Friends Do Not Just Give Verbal Advice: They Can Update Your Weights
• 6. Our response to the TanStack npm supply chain attack
• 7. GridSFM: A new, small foundation model for the electric grid

Lumen Research Digest — 2026-05-13

2026-05-13 · 6 items

• 1. LychSim: A Controllable and Interactive Simulation Framework for Vision Research
• 2. What Parameter Golf taught us about AI-assisted research
• 3. Advancing AI for materials with MatterSim: experimental synthesis, faster simulation, and multi-task models
• 4. SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecture
• 5. MEME: Multi-entity & Evolving Memory Evaluation
• 6. How NVIDIA engineers and researchers build with Codex

Lumen Research Digest — 2026-05-12

2026-05-12 · 5 items

• 1. BenchCAD: A Comprehensive, Industry-Standard Benchmark for Programmatic CAD
• 2. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests
• 3. Advancing voice intelligence with new models in the API
• 4. CADBench: A Multimodal Benchmark for AI-Assisted CAD Program Generation
• 5. From Controlled to the Wild: Evaluation of Pentesting Agents for the Real-World

Lumen Research Digest — 2026-05-11

2026-05-11 · 5 items

• 1. Reason to Play: Behavioral and Brain Alignment Between Frontier LRMs and Human Game Learners
• 2. Scaling Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber
• 3. Microsoft at NSDI 2026: Advances in large-scale networked systems
• 4. LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling
• 5. Proxy3D: Efficient 3D Representations for Vision-Language Models via Semantic Clustering and Alignment

Lumen Research Digest — 2026-05-10

2026-05-10 · 5 items

• 1. Relit-LiVE: Relight Video by Jointly Learning Environment Video
• 2. How ChatGPT learns about the world while protecting privacy
• 3. AutoAdapt: Automated domain adaptation for large language models
• 4. SkillOS: Learning Skill Curation for Self-Evolving Agents
• 5. AI CFD Scientist: Toward Open-Ended Computational Fluid Dynamics Discovery with Physics-Aware AI Agents

Lumen Research Digest — 2026-05-09

2026-05-09 · 5 items

• 1. MedHorizon: Towards Long-context Medical Video Understanding in the Wild
• 2. Running Codex safely at OpenAI
• 3. Building realistic electric transmission grid dataset at scale: a pipeline from open dataset
• 4. ActCam: Zero-Shot Joint Camera and 3D Motion Control for Video Generation
• 5. BAMI: Training-Free Bias Mitigation in GUI Grounding

Lumen Research Digest — 2026-05-08

2026-05-08 · 4 items

• 1. Parloa builds service agents customers want to talk to
• 2. ADeLe: Predicting and explaining AI performance across tasks
• 3. Simplex rethinks software development with Codex
• 4. AsgardBench: A benchmark for visually grounded interactive planning

Lumen Research Digest — 2026-05-07

2026-05-07 · 6 items

• 1. LongSeeker: Elastic Context Orchestration for Long-Horizon Search Agents
• 2. How frontier enterprises are building an AI advantage
• 3. Can we AI our way to a more sustainable world?
• 4. Executable World Models for ARC-AGI-3 in the Era of Coding Agents
• 5. Design Conductor 2.0: An agent builds a TurboQuant inference accelerator in 80 hours
• 6. Singular Bank helps bankers move fast with ChatGPT and Codex

Lumen Research Digest — 2026-05-06

2026-05-06 · 5 items

• 1. An Agent-Oriented Pluggable Experience-RAG Skill for Experience-Driven Retrieval Strategy Orchestration
• 2. Microsoft at NSDI 2026: Advances in large-scale networked systems
• 3. New ways to buy ChatGPT ads
• 4. Redefining AI Red Teaming in the Agentic Era: From Weeks to Hours
• 5. Rethinking Reasoning-Intensive Retrieval: Evaluating and Advancing Retrievers in Agentic Search Systems

Lumen Research Digest — 2026-05-05

2026-05-05 · 5 items

• 1. Standing on the Shoulders of Giants: Stabilized Knowledge Distillation for Cross--Language Code Clone Detection
• 2. OpenAI and PwC collaborate to reimagine the office of the CFO
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. FlexSQL: Flexible Exploration and Execution Make Better Text-to-SQL Agents
• 5. When Audio-Language Models Fail to Leverage Multimodal Context for Dysarthric Speech Recognition

Lumen Research Digest — 2026-05-04

2026-05-04 · 5 items

• 1. Self-Adaptive Multi-Agent LLM-Based Security Pattern Selection for IoT Systems
• 2. Our commitment to community safety
• 3. GroundedPlanBench: Spatially grounded long-horizon task planning for robot manipulation
• 4. Position: agentic AI orchestration should be Bayes-consistent
• 5. Evaluating the Architectural Reasoning Capabilities of LLM Provers via the Obfuscated Natural Number Game

Lumen Research Digest — 2026-05-03

2026-05-03 · 5 items

• 1. AEGIS: A Holistic Benchmark for Evaluating Forensic Analysis of AI-Generated Academic Images
• 2. Building the compute infrastructure for the Intelligence Age
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. FlashRT: Towards Computationally and Memory Efficient Red-Teaming for Prompt Injection and Knowledge Corruption
• 5. FlexiTac: A Low-Cost, Open-Source, Scalable Tactile Sensing Solution for Robotic Systems

Lumen Research Digest — 2026-05-02

2026-05-02 · 5 items

• 1. Exploration Hacking: Can LLMs Learn to Resist RL Training?
• 2. Where the goblins came from
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. OmniRobotHome: A Multi-Camera Platform for Real-Time Multiadic Human-Robot Interaction
• 5. PRISM: Pre-alignment via Black-box On-policy Distillation for Multimodal Reinforcement Learning

Lumen Research Digest — 2026-05-01

2026-05-01 · 5 items

• 1. Visual Generation in the New Era: An Evolution from Atomic Mapping to Agentic World Modeling
• 2. Red-teaming a network of agents: Understanding what breaks when AI agents interact at scale
• 3. Introducing Advanced Account Security
• 4. HERMES++: Toward a Unified Driving World Model for 3D Scene Understanding and Generation
• 5. Generalizable Sparse-View 3D Reconstruction from Unconstrained Images

Lumen Research Digest — 2026-04-30

2026-04-30 · 5 items

• 1. Color-Encoded Illumination for High-Speed Volumetric Scene Reconstruction
• 2. Cybersecurity in the Intelligence Age
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. Bian Que: An Agentic Framework with Flexible Skill Arrangement for Online System Operations
• 5. World2VLM: Distilling World Model Imagination into VLMs for Dynamic Spatial Reasoning

Lumen Research Digest — 2026-04-29

2026-04-29 · 5 items

• 1. Recursive Multi-Agent Systems
• 2. OpenAI models, Codex, and Managed Agents come to AWS
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. Toward Multimodal Conversational AI for Age-Related Macular Degeneration
• 5. Variational Neural Belief Parameterizations for Robust Dexterous Grasping under Multimodal Uncertainty

Lumen Research Digest — 2026-04-28

2026-04-28 · 6 items

• 1. Learning Human-Intention Priors from Large-Scale Human Demonstrations for Robotic Manipulation
• 2. An open-source spec for orchestration: Symphony
• 3. Ideas: Steering AI toward the work future we want
• 4. AgentWard: A Lifecycle Security Architecture for Autonomous AI Agents
• 5. World-R1: Reinforcing 3D Constraints for Text-to-Video Generation
• 6. Choco automates food distribution with AI agents

Lumen Research Digest — 2026-04-27

2026-04-27 · 5 items

• 1. QuantClaw: Precision Where It Matters for OpenClaw
• 2. How to get started with Codex
• 3. New Future of Work: AI is driving rapid change, uneven benefits
• 4. EV-CLIP: Efficient Visual Prompt Adaptation for CLIP in Few-shot Action Recognition under Visual Challenges
• 5. ATRS: Adaptive Trajectory Re-splitting via a Shared Neural Policy for Parallel Optimization

Lumen Research Digest — 2026-04-26

2026-04-26 · 6 items

• 1. Vista4D: Video Reshooting with 4D Point Clouds
• 2. Codex settings
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. VistaBot: View-Robust Robot Manipulation via Spatiotemporal-Aware View Synthesis
• 5. Black-Box Skill Stealing Attack from Proprietary LLM Agents: An Empirical Study
• 6. Introducing GPT-5.5

Lumen Research Digest — 2026-04-25

2026-04-25 · 6 items

• 1. Task-Driven Co-Design of Heterogeneous Multi-Robot Systems
• 2. Plugins and skills
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. Long-Horizon Manipulation via Trace-Conditioned VLA Planning
• 5. Seeing Fast and Slow: Learning the Flow of Time in Videos
• 6. Working with Codex

Lumen Research Digest — 2026-04-24

2026-04-24 · 6 items

• 1. Learning to Communicate: Toward End-to-End Optimization of Multi-Agent Language Systems
• 2. Automations
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. Context Unrolling in Omni Models
• 5. Tool Attention Is All You Need: Dynamic Tool Gating and Lazy Schema Loading for Eliminating the MCP/Tools Tax in Scalable Agentic Workflows
• 6. Top 10 uses for Codex at work

Lumen Research Digest — 2026-04-23

2026-04-23 · 6 items

• 1. DeVI: Physics-based Dexterous Human-Object Interaction via Synthetic Video Imitation
• 2. Introducing workspace agents in ChatGPT
• 3. AutoAdapt: Automated domain adaptation for large language models
• 4. Automatic Ontology Construction Using LLMs as an External Layer of Memory, Verification, and Planning for Hybrid Intelligent Systems
• 5. Where and What: Reasoning Dynamic and Implicit Preferences in Situated Conversational Recommendation
• 6. Speeding up agentic workflows with WebSockets in the Responses API

Lumen Research Digest — 2026-04-22

2026-04-22 · 5 items

• 1. A-MAR: Agent-based Multimodal Art Retrieval for Fine-Grained Artwork Understanding
• 2. Scaling Codex to enterprises worldwide
• 3. Will machines ever be intelligent?
• 4. CityRAG: Stepping Into a City via Spatially-Grounded Video Generation
• 5. SafetyALFRED: Evaluating Safety-Conscious Planning of Multimodal Large Language Models

Lumen Research Digest — 2026-04-21

2026-04-21 · 5 items

• 1. MultiWorld: Scalable Multi-Agent Multi-View Video World Models
• 2. Can we AI our way to a more sustainable world?
• 3. OpenAI helps Hyatt advance AI among colleagues
• 4. OneVL: One-Step Latent Reasoning and Planning with Vision-Language Explanation
• 5. Using large language models for embodied planning introduces systematic safety risks

Lumen Research Digest — 2026-04-20

2026-04-20 · 5 items

• 1. FineCog-Nav: Integrating Fine-grained Cognitive Modules for Zero-shot Multimodal UAV Navigation
• 2. ChatGPT for research
• 3. ADeLe: Predicting and explaining AI performance across tasks
• 4. DENALI: A Dataset Enabling Non-Line-of-Sight Spatial Reasoning with Low-Cost LiDARs
• 5. Semantic Area Graph Reasoning for Multi-Robot Language-Guided Search

Lumen Research Digest — 2026-04-19

2026-04-19 · 5 items

• 1. Blue Data Intelligence Layer: Streaming Data and Agents for Multi-source Multi-modal Data-Centric Applications
• 2. Creating images with ChatGPT
• 3. PlugMem: Transforming raw agent interactions into reusable knowledge
• 4. RadAgent: A tool-using AI agent for stepwise interpretation of chest computed tomography
• 5. Prism: Symbolic Superoptimization of Tensor Programs

Lumen Research Digest — 2026-04-18

2026-04-18 · 5 items

• 1. MM-WebAgent: A Hierarchical Multimodal Web Agent for Webpage Generation
• 2. Codex for (almost) everything
• 3. GroundedPlanBench: Spatially grounded long-horizon task planning for robot manipulation
• 4. TokenLight: Precise Lighting Control in Images using Attribute Tokens
• 5. Learning to Think Like a Cartoon Captionist: Incongruity-Resolution Supervision for Multimodal Humor Understanding

Lumen Research Digest — 2026-04-17

2026-04-17 · 5 items

• 1. GlobalSplat: Efficient Feed-Forward 3D Gaussian Splatting via Global Scene Tokens
• 2. Introducing GPT-Rosalind for life sciences research
• 3. Think in Latent Thoughts: A New Paradigm for Gloss-Free Sign Language Translation
• 4. CoopEval: Benchmarking Cooperation-Sustaining Mechanisms and LLM Agents in Social Dilemmas
• 5. Accelerating the cyber defense ecosystem that protects us all

Lumen Research Digest — 2026-04-16

2026-04-16 · 4 items

• 1. The next evolution of the Agents SDK
• 2. AsgardBench: A benchmark for visually grounded interactive planning
• 3. Enterprises power agentic workflows in Cloudflare Agent Cloud with OpenAI
• 4. Systematic debugging for AI agents: Introducing the AgentRx framework

Lumen Research Digest — 2026-04-15

2026-04-15 · 5 items

• 1. RMGS-SLAM: Real-time Multi-sensor Gaussian Splatting SLAM
• 2. Trusted access for the next era of cyber defense
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. Lyra 2.0: Explorable Generative 3D Worlds
• 5. Toward Autonomous Long-Horizon Engineering for ML Research

Lumen Research Digest — 2026-04-14

2026-04-14 · 5 items

• 1. LMMs Meet Object-Centric Vision: Understanding, Segmentation, Editing and Generation
• 2. Enterprises power agentic workflows in Cloudflare Agent Cloud with OpenAI
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. StarVLA-$α$: Reducing Complexity in Vision-Language-Action Systems
• 5. LottieGPT: Tokenizing Vector Animation for Autoregressive Generation

Lumen Research Digest — 2026-04-13

2026-04-13 · 3 items

• 1. Working with files in ChatGPT
• 2. AsgardBench: A benchmark for visually grounded interactive planning
• 3. Systematic debugging for AI agents: Introducing the AgentRx framework

Lumen Research Digest — 2026-04-12

2026-04-12 · 5 items

• 1. RewardFlow: Generate Images by Optimizing What You Reward
• 2. Our response to the Axios developer tool compromise
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. Seeing but Not Thinking: Routing Distraction in Multimodal Mixture-of-Experts
• 5. E-3DPSM: A State Machine for Event-Based Egocentric 3D Human Pose Estimation

Lumen Research Digest — 2026-04-11

2026-04-11 · 5 items

• 1. Scal3R: Scalable Test-Time Training for Large-Scale 3D Reconstruction
• 2. Applications of AI at OpenAI
• 3. AsgardBench: A benchmark for visually grounded interactive planning
• 4. OpenVLThinkerV2: A Generalist Multimodal Reasoning Model for Multi-domain Visual Tasks
• 5. PSI: Shared State as the Missing Layer for Coherent AI-Generated Instruments in Personal AI Agents

Lumen Research Digest — 2026-04-10

2026-04-10 · 7 items

• 1. Visually-grounded Humanoid Agents
• 2. CyberAgent moves faster with ChatGPT Enterprise and Codex
• 3. New Future of Work: AI is driving rapid change, uneven benefits
• 4. Act Wisely: Cultivating Meta-Cognitive Tool Use in Agentic Multimodal Models
• 5. AVGen-Bench: A Task-Driven Benchmark for Multi-Granular Evaluation of Text-to-Audio-Video Generation
• 6. Ideas: Steering AI toward the work future we want
• 7. OpenAI Full Fan Mode Contest: Terms & Conditions

Lumen Research Digest — 2026-04-09

2026-04-09 · 4 items

• 1. Introducing the Child Safety Blueprint
• 2. Phi-4-reasoning-vision and the lessons of training a multimodal reasoning model
• 3. The next phase of enterprise AI
• 4. AsgardBench: A benchmark for visually grounded interactive planning

Lumen Research Digest — 2026-04-08

2026-04-08 · 5 items

• 1. CritBench: A Framework for Evaluating Cybersecurity Capabilities of Large Language Models in IEC 61850 Digital Substation Environments
• 2. Announcing the OpenAI Safety Fellowship
• 3. Phi-4-reasoning-vision and the lessons of training a multimodal reasoning model
• 4. CoStream: Codec-Guided Resource-Efficient System for Video Streaming Analytics
• 5. MMEmb-R1: Reasoning-Enhanced Multimodal Embedding with Pair-Aware Selection and Adaptive Control

Lumen Research Digest — 2026-04-07

2026-04-07 · 5 items

• 1. FileGram: Grounding Agent Personalization in File-System Behavioral Traces
• 2. Industrial policy for the Intelligence Age
• 3. Phi-4-reasoning-vision and the lessons of training a multimodal reasoning model
• 4. Analyzing Symbolic Properties for DRL Agents in Systems and Networking
• 5. QED-Nano: Teaching a Tiny Model to Prove Hard Theorems

Lumen Research Digest — 2026-04-06

2026-04-06 · 5 items

• 1. Coupled Control, Structured Memory, and Verifiable Action in Agentic AI (SCRAT -- Stochastic Control with Retrieval and Auditable Trajectories): A Comparative Perspective from Squirrel Locomotion and Scatter-Hoarding
• 2. Helping disaster response teams turn AI into action across Asia
• 3. Phi-4-reasoning-vision and the lessons of training a multimodal reasoning model
• 4. FSUNav: A Cerebrum-Cerebellum Architecture for Fast, Safe, and Universal Zero-Shot Goal-Oriented Navigation
• 5. A Systematic Security Evaluation of OpenClaw and Its Variants

Lumen Research Digest — 2026-04-05

2026-04-05 · 5 items

• 1. Modulate-and-Map: Crossmodal Feature Mapping with Cross-View Modulation for 3D Anomaly Detection
• 2. STADLER reshapes knowledge work at a 230-year-old company
• 3. Phi-4-reasoning-vision and the lessons of training a multimodal reasoning model
• 4. Batched Contextual Reinforcement: A Task-Scaling Law for Efficient Reasoning
• 5. Model-Based Reinforcement Learning for Control under Time-Varying Dynamics

Lumen Research Digest — 2026-04-04

2026-04-04 · 5 items

• 1. SKILL0: In-Context Agentic Reinforcement Learning for Skill Internalization
• 2. OpenAI acquires TBPN
• 3. Phi-4-reasoning-vision and the lessons of training a multimodal reasoning model
• 4. Large-scale Codec Avatars: The Unreasonable Effectiveness of Large-scale Avatar Pretraining
• 5. Generative World Renderer

Lumen Research Digest — 2026-04-03

2026-04-03 · 5 items

• 1. ActionParty: Multi-Subject Action Binding in Generative Video Games
• 2. Codex now offers more flexible pricing for teams
• 3. Trailer: The Shape of Things to Come
• 4. Stop Wandering: Efficient Vision-Language Navigation via Metacognitive Reasoning
• 5. VOID: Video Object and Interaction Deletion

Lumen Research Digest — 2026-04-02

2026-04-02 · 4 items

• 1. Gradient Labs gives every bank customer an AI account manager
• 2. ADeLe: Predicting and explaining AI performance across tasks
• 3. Will machines ever be intelligent?
• 4. Introducing the OpenAI Safety Bug Bounty program

Lumen Research Digest — 2026-04-01

2026-04-01 · 6 items

• 1. SurgTEMP: Temporal-Aware Surgical Video Question Answering with Text-guided Visual Memory for Laparoscopic Cholecystectomy
• 2. PlugMem: Transforming raw agent interactions into reusable knowledge
• 3. Accelerating the next phase of AI
• 4. EC-Bench: Enumeration and Counting Benchmark for Ultra-Long Videos
• 5. HapCompass: A Rotational Haptic Device for Contact-Rich Robotic Teleoperation
• 6. CORPGEN advances AI agents for real work

Lumen Research Digest — 2026-03-31

2026-03-31 · 4 items

• 1. Systematic debugging for AI agents: Introducing the AgentRx framework
• 2. Update on the OpenAI Foundation
• 3. GroundedPlanBench: Spatially grounded long-horizon task planning for robot manipulation
• 4. Creating with Sora Safely

Lumen Research Digest — 2026-03-30

2026-03-30 · 6 items

• 1. Beyond Language: Grounding Referring Expressions with Hand Pointing in Egocentric Vision
• 2. Phi-4-reasoning-vision and the lessons of training a multimodal reasoning model
• 3. Creating with Sora Safely
• 4. GaussianGPT: Towards Autoregressive 3D Gaussian Scene Generation
• 5. Drive-Through 3D Vehicle Exterior Reconstruction via Dynamic-Scene SfM and Distortion-Aware Gaussian Splatting
• 6. AsgardBench: A benchmark for visually grounded interactive planning

Lumen Research Digest — 2026-03-29

2026-03-29 · 7 items

• 1. Introducing the OpenAI Safety Bug Bounty program
• 2. Powering product discovery in ChatGPT
• 3. A New Framework for Evaluating Voice Agents (EVA)
• 4. Holotron-12B - High Throughput Computer Use Agent
• 5. How we monitor internal coding agents for misalignment
• 6. Inside our approach to the Model Spec
• 7. Helping developers build safer AI experiences for teens

Lumen Research Digest — 2026-03-28

2026-03-28 · 7 items

• 1. Vega: Learning to Drive with Natural Language Instructions
• 2. Persistent Robot World Models: Stabilizing Multi-Step Rollouts via Reinforcement Learning
• 3. Back to Basics: Revisiting ASR in the Age of Voice Agents
• 4. The Kitchen Loop: User-Spec-Driven Development for a Self-Evolving Codebase
• 5. Natural-Language Agent Harnesses
• 6. Agent Factories for High Level Synthesis: How Far Can General-Purpose Coding Agents Go in Hardware Optimization?
• 7. Drive My Way: Preference Alignment of Vision-Language-Action Model for Personalized Driving

Lumen Research Digest — 2026-03-27

2026-03-27 · 7 items

• 1. Less Gaussians, Texture More: 4K Feed-Forward Textured Splatting
• 2. Vega: Learning to Drive with Natural Language Instructions
• 3. Is Mathematical Problem-Solving Expertise in Large Language Models Associated with Assessment Performance?
• 4. LanteRn: Latent Visual Structured Reasoning
• 5. R-C2: Cycle-Consistent Reinforcement Learning Improves Multimodal Reasoning
• 6. Missing-Aware Multimodal Fusion for Unified Microservice Incident Management
• 7. RefAlign: Representation Alignment for Reference-to-Video Generation