| JevGPT: Thinking Fast and Slow in Autonomous Research | ✓ Link | 90.48 | JevGPT | 2026-10-13 |
| Introducing Claude Opus 4.5 | | 85.30 | Claude Opus 4.5 (w/ tools) | 2025-11-24 |
| Masking Stale Observations Helps Search Agents -- Until It Doesn't: A Regime Map and Its Mechanism | ✓ Link | 84.3 | DS-V4-Flash-Max + AgentIR-4B (stale-observation masking, K=5) | 2026-05-29 |
| Rethinking Agentic Search with Pi-Serini: Is Lexical Retrieval Sufficient? | ✓ Link | 83.1 | PI-SERINI (gpt-5.5) | 2026-05-11 |
| JIT-Agent: Scaling Harness Intelligence via Just-in-Time Harness Evolution | ✓ Link | 78.0 | JIT-Agent + GLM-5.2 | 2026-08-26 |
| Agentic Aggregation for Parallel Scaling of Long-Horizon Agentic Tasks | ✓ Link | 77.33 | AggAgent (MiniMax-M2.5) | 2026-04-13 |
| Self-Evolving Search Index | ✓ Link | 76.02 | Self-Index (Gemini-3.7-Flash, BM25) | 2026-09-17 |
| TacoMAS: Test-Time Co-Evolution of Topology and Capability in LLM-based Multi-Agent Systems | ✓ Link | 74.5 | TacoMAS | 2026-05-10 |
| Adaptive Consistency Graph for Long-Horizon Agents | ✓ Link | 73.5 | ACG (GPT-5.6-luna) | 2026-09-26 |
| Dr-DCI: Scaling Direct Corpus Interaction via Dynamic Workspace Expansion | ✓ Link | 73.25 | DR-DCI + Workspace-Preserving Context Reset | 2026-06-12 |
| KARL: Knowledge Agents via Reinforcement Learning | | 70.4 | KARL-BCP (VGS N=17) | 2026-03-05 |
| QUEST: Training Frontier Deep Research Agents with Fully Synthetic Tasks | ✓ Link | 69.5 | QUEST-35B | 2026-05-22 |
| FABLE: Forest-Based Adaptive Bi-Path LLM-Enhanced Retrieval for Multi-Document Reasoning | | 66.6 | TDR-30B-A3B (with FABLE retriever) | 2026-01-26 |
| Deep Search with Hierarchical Meta-Cognitive Monitoring Inspired by Cognitive Neuroscience | | 62.0 | DS-MCM (with Tongyi-DeepResearch backbone; memory=BrowseComp) | 2026-01-30 |
| DAGent: Evaluate-then-Grow Planning for Deep Research Agents | ✓ Link | 61.3 | DAGent (Qwen3-235B-A22B) | 2026-09-30 |
| Context Language Models | ✓ Link | 59.4 | CLM (Qwen3.6-27B) | 2026-09-29 |
| ContextPilot: Teaching Agents for Proactive Context Management via Fine-grained RL | ✓ Link | 55.5 | ContextPilot-14B-RL | 2026-08-28 |
| OpenResearcher: A Fully Open Pipeline for Long-Horizon Deep Research Trajectory Synthesis | ✓ Link | 54.8 | OpenResearcher-30B-A3B | 2026-03-17 |
| SearchJev: A Fast and Calibrated System-1 Model for Search Agents | ✓ Link | 54.0 | SearchJev-4B | 2026-10-04 |
| LACUNA: Safe Agents as Recursive Program Holes | ✓ Link | 27.1 | Lacuna (DeepSeek-V4-Flash) | 2026-05-27 |
| LFM2.5-2.6B: Deploy Agents Everywhere | ✓ Link | 26.89 | LFM2.5-2.6B (OpenClaw) | 2026-08-04 |
| WideSeek: Advancing Wide Research via Multi-Agent Scaling | ✓ Link | 26.42 | WideSeek-8B-RL | 2026-02-02 |