| DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression | ✓ Link | 68.2 | DeepSeek-V4.1-Flash-1017 | 2026-09-17 |
| DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression | ✓ Link | 65.4 | DeepSeek-V4.1-Flash Max | 2026-09-17 |
| DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence | ✓ Link | 61.5 | DeepSeek-V4-Pro-0813 (Max) | 2026-04-26 |
| Hy4 preview | ✓ Link | 58.9 | Hy4 preview | 2026-08-28 |
| GLM-5.3: Frontier Coding with Emergent Cyber Capabilities | ✓ Link | 58.0 | GLM-5.3 | 2026-08-14 |
| On the Limitations of Non-GPU AI Accelerators for Large-Model Inference: A Field Study of MoE and Multimodal Serving on Huawei Ascend | | 57.7 | DeepSeek-V4-Flash-Vision-Exp | 2026-07-09 |
| GLM-5.3-Flash: Frontier Intelligence, Flash Cost | ✓ Link | 56.3 | GLM-5.3-Flash | 2026-08-26 |
| Qwen3.8-Max: A New Bar for Coding and Cowork | ✓ Link | 55.9 | Qwen3.8-Max | 2026-08-03 |
| DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence | ✓ Link | 54.2 | DeepSeek-V4-Flash-0731 (Max) | 2026-04-26 |
| Introducing GPT-5.5 | | 50.7 | GPT-5.5 (Max) | 2026-04-23 |
| dots3-note Preview: A Small but Mighty Step Toward Long-Horizon Agency in Real Life | ✓ Link | 49.82 | dots3-note Preview | 2026-08-13 |
| GLM-5.2: Built for Long-Horizon Tasks | ✓ Link | 48.9 | GLM-5.2 (Max) | 2026-06-16 |
| Ornith-1.0: Self-Scaffolding LLMs for Agentic Coding | ✓ Link | 48.2 | Ornith-1.0-397B | 2026-06-21 |
| On the Design of Qwen3.8-Next Architecture: Evaluation, Efficiency, and Training Stability | ✓ Link | 48.1 | Qwen3.8-Flash-Next | 2026-08-31 |
| Seed2.1 Model Card: Agentic Intelligence for Productivity | | 47.0 | Seed2.1 Pro | 2026-06-23 |
| Ornith-1.5: From Self-Scaffolding to Self-Improvement | | 46.2 | Ornith-1.5-35B-A3B | 2026-08-20 |
| Hy3 | ✓ Link | 45.6 | Hy3 | 2026-07-02 |
| GLM-5.1: Towards Long-Horizon Tasks | ✓ Link | 42.7 | GLM-5.1 (Max) | 2026-04-03 |
| Qwen3.8-27B | ✓ Link | 42.3 | Qwen3.8-27B | 2026-08-05 |
| The MiniMax-M2 Series: Mini Activations Unleashing Max Real-World Intelligence | ✓ Link | 39.8 | MiniMax-M2.7 (229.9B total, 9.8B activated) | 2026-05-26 |
| Marathoner: Ultra-Long-Horizon Autonomous Intelligence | | 34.7 | Marathoner-9B | 2026-09-28 |
| Ornith-1.0: Self-Scaffolding LLMs for Agentic Coding | ✓ Link | 34.6 | Ornith-1.0-35B | 2026-06-21 |
| Gemini 3.1 Pro: A smarter model for your most complex tasks | | 33.4 | Gemini 3.1 Pro (Max) | 2026-02-19 |
| Ornith-1.0: Self-Scaffolding LLMs for Agentic Coding | ✓ Link | 27.2 | Ornith-1.0-9B | 2026-06-21 |