| GLM-5: From Vibe Coding to Agentic Engineering | ✓ Link | 89.7 | GLM-5 | 2026-02-11 |
| Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters | ✓ Link | 88.2 | Step-3.5-Flash | 2026-02-11 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 86.7 | Qwen3.5-397B-A17B | 2026-02-16 |
| Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale | ✓ Link | 84.26 | Ring-2.6-1T (high) | 2026-06-13 |
| Kwai Keye-VL-2.0 Technical Report | ✓ Link | 82.6 | Keye-VL-2.0-30B-A3B | 2026-06-09 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 81.2 | Qwen3.5-35B-A3B | 2026-02-16 |
| Intern-S1-Pro: Scientific Multimodal Foundation Model at Trillion Scale | | 80.9 | Intern-S1-Pro | 2026-03-26 |
| DeepSeek-V3.2 Technical Report | ✓ Link | 80.3 | DeepSeek-V3.2 | 2025-12-02 |
| MiMo-V2-Flash Technical Report | ✓ Link | 80.3 | MiMo-V2-Flash (Thinking) | 2026-01-06 |
| ToolOrchestra: Elevating Intelligence via Efficient Model and Tool Orchestration | ✓ Link | 80.2 | Orchestrator | 2025-11-26 |
| Mach-Mind-4-Flash Technical Report | | 80.04 | Mach-Mind-4-Flash | 2026-07-10 |
| Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent | ✓ Link | 79.8 | Agents-A1 | 2026-06-29 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 79.5 | Qwen3.5-122B-A10B | 2026-02-16 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 79.1 | Qwen3.5-9B | 2026-02-16 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 79.0 | Qwen3.5-27B | 2026-02-16 |
| ERNIE 5.0 Technical Report | | 78.79 | ERNIE 5.0 | 2026-02-04 |
| Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale | ✓ Link | 78.36 | Ling-2.6-1T | 2026-06-13 |
| Gemma 4 Technical Report | ✓ Link | 76.9 | Gemma 4 31B | 2026-07-02 |
| Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale | ✓ Link | 76.36 | Ling-2.6-flash | 2026-06-13 |
| LACUNA: Safe Agents as Recursive Program Holes | ✓ Link | 76.0 | Lacuna (deepseek-v4-flash) | 2026-05-27 |
| SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecture | ✓ Link | 75.39 | SenseNova-U1-A3B-MoT | 2026-05-12 |
| WEFT: Scaling Tool-Use Post-Training for General-Purpose Agents | | 74.34 | WEFT-35B-A3B | 2026-09-29 |
| Announcing Bonsai 27B: The First 27B-Class Model to Run on a Phone | ✓ Link | 73.61 | Ternary Bonsai 27B | 2026-07-14 |
| EXAONE 4.5 Technical Report | ✓ Link | 72.0 | EXAONE 4.5 (Reasoning) | 2026-04-09 |
| Gemma 4 Technical Report | ✓ Link | 69.0 | Gemma 4 12B Unified | 2026-07-02 |
| Kimi K2: Open Agentic Intelligence | ✓ Link | 66.1 | Kimi K2 | 2025-07-28 |
| Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence | | 65.4 | Agent-World (14B) | 2026-04-20 |
| PivotRL: High Accuracy Agentic Post-Training at Low Compute Cost | | 63.81 | Qwen3-30B-A3B-Thinking-2507 + PivotRL | 2026-03-22 |
| ASTRA: Automated Synthesis of agentic Trajectories and Reinforcement Arenas | ✓ Link | 63.7 | ASTRA-32B-thinking-v1 | 2026-01-29 |
| NVIDIA Nemotron 3: Efficient and Open Intelligence | ✓ Link | 61.15 | Nemotron 3 Super | 2025-12-24 |
| Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation | | 58.9 | Nemotron-Cascade-2-30B-A3B | 2026-03-19 |
| Unified Audio Intelligence Without Regressing on Text Intelligence | | 57.2 | Audex 30B-A3B | 2026-07-06 |
| DiffusionGemma Technical Report | ✓ Link | 56.2 | DiffusionGemma 26B A4B | 2026-07-31 |
| Nanbeige4.1-3B: A Small General Model that Reasons, Aligns, and Acts | ✓ Link | 48.57 | Nanbeige4.1-3B | 2026-02-13 |
| Reinforcement World Model Learning for LLM-based Agents | | 43.7 | RWML + Policy RL (Qwen3-8B) | 2026-02-05 |
| ARC: Fair Relative Advantage Comparison in Open-Ended Real-World Interaction | | 38.22 | GRPO + ARC (Qwen3-8B) | 2026-08-13 |
| From Production Traffic to Post-Training: Building a Self-Hosted LLM That Covers the Corporate Request Mix | | 37.6 | T-pro2.1 (internal, non-reasoning) | 2026-09-01 |
| EnvACE: Internalizing Environment Dynamics via World Rehearsal for Agentic Reinforcement Learning | ✓ Link | 36.7 | EnvACE-8B | 2026-08-06 |
| EnvFactory: Scaling Tool-Use Agents via Executable Environments Synthesis and Robust RL | ✓ Link | 33.67 | Qwen3-8B (EnvFactory) | 2026-05-18 |
| SEAL: Synergistic Co-Evolution of Agents and Learning Environments | ✓ Link | 20.79 | Qwen2.5-7B-Instruct (SEAL) | 2026-05-23 |
| Youtu-LLM: Unlocking the Native Agentic Potential for Lightweight Large Language Models | ✓ Link | 15.0 | Youtu-LLM 2B | 2025-12-31 |