Frontier AI PM Radar (raw) — 2026-10-21¶
LLM synthesis was unavailable for this run; these are the top pre-ranked items.
- Industrial ReMEmbR: Scaling Spatiotemporal Memory to Robot Fleets · arXiv · score 8.851
- WorldSR: Harnessing World Knowledge Search for Grounded Image Super-Resolution · arXiv · score 8.851
- Announcing MCP support in Apigee hybrid: agentic tools that run in your own cluster · Google Developers · score 8.8
- TraceLite: Efficient Trace Reduction for SWE Agents via Optical Compression and Redundancy Removal · arXiv · score 8.785
- S-Agent 2.0: Spatial Reasoning with a General Coding Agent · arXiv · score 8.785
- DeclineCraft: A RAG Abstention Benchmark for Jointly Measuring Combined Retrieval and Parametric Knowledge · arXiv · score 8.785
- EvaOmni: An Evidence-Grounded Benchmark for Audio-Visual Reasoning · HF Daily Papers · score 8.716
- Breaking Diffusion: Parallel Reasoning with Draft Refinement · HF Daily Papers · score 8.663
- When Do LLMs Help with Feature Selection? A Systematic Empirical Study · arXiv · score 8.613
- After the Attacker Leaves: Benchmarking Carryover Attacks on LLM Agents · arXiv · score 8.613
- Parameter-Efficient Adaptation of Time Series Foundation Models for Electrical Load Forecasting · arXiv · score 8.613
- Self-Credit: Credit Assignment for Long-Horizon Agentic RL via Intra-group Rescaling · arXiv · score 8.491
- Object-Oriented Harness: Evolving Agent Harnesses through Object-Oriented Design · arXiv · score 8.491
- GALA: Geometry-Aware Language Model for Controllable Object Arrangement · arXiv · score 8.491
- EdgeKernelBench: Unified LLM-Driven Kernel Generation for Edge Backends · HF Daily Papers · score 8.491
- More Memory, Worse Agents: Error Reproduction and Anti-Persistence in LLM Agents · HF Daily Papers · score 8.342
- HarnessRL: Co-Evolving Policies with a Just-in-Time Training Harness for Open-Ended RL · HF Daily Papers · score 8.279
- Patch the Planet: global response coordination as agents uncover critical vulnerabilities · OpenAI News · score 8.25
- Matrix-Game 4.0: Harnessing Video Models for Real-Time Playable Game Worlds · HF Daily Papers · score 8.219
- HybridResearcher: Bridging the Structured-Unstructured Boundary for Deep Research Agents · HF Daily Papers · score 8.204
- Muse Glimmer 30B: An On-Device Agentic Model · HF Daily Papers · score 8.114
- Graphicle-AI/graphicle · GitHub · score 8.026
- MAS-Guard: Unified Guardrails from Individual Agents to Multi-Agent LLM Systems · arXiv · score 7.851
- Leonxlnx/earshot · GitHub · score 7.809
- FairStress: A deeper evaluation of fairness in LLM decision-making · Hugging Face Blog · score 7.8
- ctoldigital/axuse · GitHub · score 7.788
- What Do RTL Repair Benchmarks Actually Measure? · arXiv · score 7.785
- Agentic Ensemble with Vision-Language Models for Visual Anomaly Detection · arXiv · score 7.785
- DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems · arXiv · score 7.785
- OptScientist: Automated Discovery of Optimization Laws · arXiv · score 7.708
- CAPSafe: A Benchmark for Contextual Action and Plan Safety Reasoning in Embodied Agents · arXiv · score 7.708
- Announcing New vLLM and SGLang Inference Engines on TPU · Google Developers · score 7.7
- Three layers, one repo: a tour of google/adk-recipes · Google Developers · score 7.7
- Supercharge your development with the Google Developer Knowledge API ecosystem · Google Developers · score 7.7
- SpecKCast: Spectral Kalman Forecasting for Diffusion Acceleration · HF Daily Papers · score 7.633
- GBClaw: An agentic AI framework for atomistic grain boundary research and mechanistic discovery · arXiv · score 7.613
- Benchmarking Arabic Dialects for Math Problems · arXiv · score 7.613
- Value-Rich, Differently Organized: Measuring Value Expression and Organization in AI-Generated Debates · arXiv · score 7.613
- Capable Teacher, Myopic Scorer: Reward-Free Teacher Repair for Agentic On-Policy Distillation · arXiv · score 7.613
- FALCA: Falsification-Aware Reinforcement Learning with Hierarchical Credit Assignment for Long-Horizon Deep Search Agents · arXiv · score 7.613
Scanned 371 items from 28 sources.