Archived
2026-09-30Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning
Paras Dahal, Anton Bakhtin, Taco Cohen, Zhengxing Chen, Carole-Jean Wu, Rob Fergus, et al.
arXiv:2609.38147v1When AI agents tackle longer, messier problems, the hard part is no longer just solving the task — it’s deciding how to spend the next step of computation. This paper introduces agentic meta-reasoning, a controller that sits above the worker model and explicitly reasons about control: what partial work to reuse, when to branch into a new attempt, when to stop, and how to allocate the remaining budget. Instead of replaying the full history every time, the controller keeps a compact summary of what has been learned so far and uses that to choose the next move. The result is a more structured way to run agents on long-horizon tasks. On program reconstruction, abstract reasoning, and proof generation, the method beats direct control and keeps improving where simpler approaches plateau. Why it matters is simple: as agents get more capable, better orchestration can unlock gains that raw model size alone cannot, making inference-time planning a key part of scaling AI.
Also spotted that day
Skill-Space Shooting for Autonomous Robot Policy ImprovementImagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before AnsweringBreakdown of Local Denoising as Semantic SpeciationSTEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State QuantizationLeapQuant: Efficient Linear Attention with Accurate Recurrent State QuantizationCropland PAtteRNS: Parallel Dimensional Attention Networks and Attention to Dataset Disparity for Crop Segmentation in Satellite Imagery Time Series DataA Spectral Theory of Distortion in LLM Graph Reconstruction: Sharp Bounds and Empirical CharacterizationEmoRES-TTS: Residual-Enhanced Vector Steering for Emotional Speech GenerationBeyond the Timeline: Augmenting Long-Video Memory with Grounded Entity BiographiesPretraining Latent Information Feedback Transformers with Teacher SupervisionLearning Meta-Skills for Agent Harness Design in Test-Time AI4AIAdviSD: Learning to Advise Frontier LLMs via Targeted Multi-Turn Self-DistillationBreaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoELongHarness Bench: Stress-Testing Language Model Harnesses for Long-Context Reasoning
Previous daily papers
2026-10-15Latent Compaction with Test-Time Training2026-10-14Tev: Scaling Long-Horizon Agentic Environments with Dense Progress Rewards2026-10-14The Last Remote Sensing Model Built by Humans? Casting Any Earth Observation Task as Recursive Self-Improvement2026-10-14Hybrid Reasoning Agents2026-10-13AI Models Can Provably Hide Arbitrary Capabilities2026-10-13RATED: Reliability-Aware AI Agent Tool Retrieval via Execution Dynamics2026-10-13Designing Protein Structures from Natural Language2026-10-12RSI-Anything: Unlocking Recursive Self-Improvement with World Models2026-10-11A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control2026-10-11FAITH: Feasibility-Aware Safety-Filtered RL for High-Dimensional Systems2026-10-11VioLA: Learning Generalist Humanoid Control Policies from Human Data2026-10-10FastBench: Can Streaming VLMs Perceive High-Dynamic Real-World Streams?2026-10-10Ecology of AI Agents: Collaboration Creates a Population Threshold for Takeoff2026-10-08Long-WAM: Scaling the Context of World-Action Models2026-10-08SciExam for ENSO: Can AI Agents Build Climate Models?2026-10-08RoboJEPA: Scaling Robotic Latent World Models2026-10-08RECAST: Learning to Compute the Right Context through Adaptive Evidence Routing2026-10-07WorldSonus: Bringing Sound to Worlds2026-10-07IdeaAnchor: Teaching LLMs to Turn Literature into Research Ideas2026-10-07Agent in a Bottle: Can LLM Agents Turn Their Capabilities Into Cheap, Scalable Artifacts?2026-10-06CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling2026-10-06PlotGround: Grounding Plot Digitization in Real Scientific Figures and Their Source Data2026-10-06Base Models Can Reason By Taking a Cue From Training Data2026-10-05LESSER: Post-Training Data Selection with Output-Layer Gradients2026-10-05Language Models that Play Chess and Explain Their Moves2026-10-05Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis2026-10-04One Basis to Animate Them All: Gaussian Blendshape Distillation for Real-Time Avatars2026-10-04The Missing Primitive: Diagnosing and Repairing Mathematical Reasoning in Large Language Models2026-10-04DMAD: Distribution Matching as Adversarial Distillation for Fast Visual Generation2026-10-03VISTA: A Visual Harness for Reasoning in an Interactive World2026-10-01Looped Diffusion Transformer2026-10-01Scaling Laws for Looped Mixture of Experts2026-10-01EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery2026-09-30Beyond the Timeline: Augmenting Long-Video Memory with Grounded Entity Biographies2026-09-30Pretraining Latent Information Feedback Transformers with Teacher Supervision2026-09-29Unifying Distributional Training for One-Step Visual Generation2026-09-29Improving Test-Time Scaling with Adaptive Looped Transformers2026-09-29Telescopic Language Models2026-09-28Statistical attribute alignment for black-box generative AI via output post-processing2026-09-28User Model Extraction via Belief Self-Distillation2026-09-28Learning to Stop without Learning to Stop: Self-Supervised Confidence Training Improves Reasoning Efficiency2026-09-27TrackEverything: Long Horizon Dense Tracking via De-Duplicating 3D Scene Representations2026-09-27LLM Agents Can Easily Tamper With Their Own Traces2026-09-27Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure2026-09-26PoEM: Predicting RL Outcomes from Existing Policies2026-09-24Frozen Flows Forget: Diagnosing and Restoring Lost Motion in a Latent-flow World Model2026-09-24StudentBench: AI and human tutoring yield equivalent GRE learning gains2026-09-24Agent-Editing World Model: Rethinking World Modeling for LLM Agents2026-09-23Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents2026-09-23FleXray: Universal Clinical X-ray Segmentation2026-09-23SWE-Serve: Benchmarking Agentic Engineering For Production Inference Serving2026-09-22GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay2026-09-22Emergent Collusion in Long-Horizon LLM Agent Interaction2026-09-22WorldCrafter: Consistent Video World Model with Implicit 3D-aware Memory2026-09-21Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw2026-09-21Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design2026-09-21CodeMidas: Scaling Agentic Coding RL Environments from Code Itself2026-09-20RetireOPD: Self-Retiring On-Policy Distillation for Agentic Reinforcement Learning2026-09-20Quantifying Overclaiming Propensity in Frontier LLM Agents