| Claude Fable 5 and Claude Mythos 5 | ✓ Link | 59.0 | Claude Mythos 5 | 2026-06-09 |
| Gemini 3.1 Pro: A smarter model for your most complex tasks | | 45.0 | Gemini 3.1 Pro | 2026-02-19 |
| Kimi K3: Open Frontier Intelligence | ✓ Link | 43.5 | Kimi K3 (max) | 2026-07-27 |
| Introducing GPT-5.4 | | 42.7 | GPT-5.4 Pro | 2026-03-05 |
| Introducing GPT-5.5 | | 41.4 | GPT-5.5 | 2026-04-23 |
| A new era of intelligence with Gemini 3 | | 41.0 | Gemini 3 Deep Think | 2025-11-18 |
| GLM-5.2: Built for Long-Horizon Tasks | ✓ Link | 40.5 | GLM-5.2 | 2026-06-16 |
| DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence | ✓ Link | 37.7 | DeepSeek-V4-Pro Max | 2026-04-26 |
| Motif 3: Technical Report | ✓ Link | 37.0 | Motif 3 | 2026-08-10 |
| Introducing GPT-5.2 | | 36.6 | GPT-5.2 Pro | 2025-12-11 |
| Introducing Beam: Reflection's 501B open-weight model | | 36.2 | Beam | 2026-10-05 |
| Qwen3.8-Flash-Next | ✓ Link | 35.9 | Qwen3.8-Flash-Next | 2026-08-24 |
| DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence | ✓ Link | 34.8 | DeepSeek-V4-Flash Max | 2026-04-26 |
| Kimi K2.6 | ✓ Link | 34.7 | Kimi K2.6 | 2026-04-20 |
| Ornith-1.5: From Self-Scaffolding to Self-Improvement | | 33.4 | Ornith-1.5-35B-A3B | 2026-08-20 |
| GLM-5.1: Towards Long-Horizon Tasks | ✓ Link | 31.0 | GLM-5.1 | 2026-04-03 |
| Qwen3.8-27B | ✓ Link | 30.8 | Qwen3.8-27B | 2026-08-05 |
| Efficient Agentic Reasoning Through Self-Regulated Simulative Planning | ✓ Link | 30.4 | SR 2 AM-v1.0-30B | 2026-05-21 |
| Kimi K2.5: Visual Agentic Intelligence | ✓ Link | 30.1 | Kimi K2.5 | 2026-02-02 |
| Hy3 preview | ✓ Link | 30.0 | Hy3 preview | 2026-04-23 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 28.7 | Qwen3.5-397B-A17B | 2026-02-16 |
| The MiniMax-M2 Series: Mini Activations Unleashing Max Real-World Intelligence | ✓ Link | 28.0 | MiniMax-M2.7 | 2026-05-26 |
| Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters | ✓ Link | 27.9 | Step-3.5-Flash PaCoRe | 2026-02-11 |
| Streaming Communication in Multi-Agent Reasoning | ✓ Link | 27.68 | STREAM-MA (Claude Opus 4.6, Graph) | 2026-06-03 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 25.3 | Qwen3.5-122B-A10B (w/ CoT) | 2026-02-16 |
| GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models | ✓ Link | 24.8 | GLM-4.7 | 2025-08-08 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 24.3 | Qwen3.5-27B (w/ CoT) | 2026-02-16 |
| Qwen3.6 | ✓ Link | 24.0 | Qwen3.6-27B | 2026-04-21 |
| Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters | ✓ Link | 23.1 | Step-3.5-Flash | 2026-02-11 |
| Kwai Keye-VL-2.0 Technical Report | ✓ Link | 22.9 | Keye-VL-2.0-30B-A3B (w/ CoT) | 2026-06-09 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 22.4 | Qwen3.5-35B-A3B (w/ CoT) | 2026-02-16 |
| Muse Glimmer 30B: An On-Device Agentic Model | ✓ Link | 22.0 | Muse Glimmer-30B (High Reasoning, text-only) | 2026-10-17 |
| Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities | | 21.6 | Gemini 2.5 Pro | 2025-07-07 |
| Kolibri: A Sovereign European Model on the Pareto Frontier | ✓ Link | 21.5 | Kolibri | 2026-10-03 |
| Qwen3.6 | ✓ Link | 21.4 | Qwen3.6-35B-A3B | 2026-04-21 |
| Introducing OpenAI o3 and o4-mini | ✓ Link | 20.32 | OpenAI o3 | 2025-04-16 |
| MiniMax-M2.5 | ✓ Link | 19.4 | MiniMax-M2.5 | 2026-02-12 |
| Intern-S2-Mobius: Foundation Model with Decoupled Knowledge and Reasoning | ✓ Link | 19.11 | Intern-S2-Mobius-35B | 2026-08-14 |
| Introducing K2 Horizon: Frontier Performance, Radically Open | ✓ Link | 18.6 | K2-Horizon-7B | 2026-09-03 |
| DIVE: Scaling Diversity in Agentic Task Synthesis for Generalizable Tool Use | ✓ Link | 17.8 | Dive-8B (RL) | 2026-03-10 |
| Nanbeige4.2-3B: Unlocking Agentic Capabilities in a Compact Model | ✓ Link | 17.8 | Nanbeige4.2-3B (w/o search) | 2026-07-21 |
| Deeper is Not Always Better: Mitigating the Alignment Tax via Confident Layer Decoding | ✓ Link | 16.8 | Confident Decoding (Qwen3.5-27B) | 2026-06-20 |
| Scalable Visual Pretraining for Language Intelligence | | 16.67 | Qwen3.5-35B-A3B + VP | 2026-07-10 |
| INTELLECT-3: Technical Report | ✓ Link | 14.6 | INTELLECT-3 | 2025-12-18 |
| GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models | ✓ Link | 14.4 | GLM-4.5 (355B) | 2025-08-08 |
| Thomson: Continual Learning of Frontier Models for SovereignAI | | 13.4 | Thomson-1.0-Small | 2026-08-24 |
| NVIDIA Nemotron 3.5 Lightning | ✓ Link | 11.72 | NVIDIA Nemotron 3.5 Lightning 30B-A3B | 2026-08-11 |
| Falcon-H1R: Pushing the Reasoning Frontiers with a Hybrid Model for Efficient Test-Time Scaling | ✓ Link | 11.1 | Falcon-H1R (7B) | 2026-01-05 |
| Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities | | 11.0 | Gemini 2.5 Flash | 2025-07-07 |
| DiffusionGemma Technical Report | ✓ Link | 11.0 | DiffusionGemma 26B A4B | 2026-07-31 |
| GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models | ✓ Link | 10.6 | GLM-4.5-Air | 2025-08-08 |
| Is Next-Chunk Reasoning RL Really Better than SFT? Revisiting Training Strategies under no-CoT Data | | 9.24 | Mixed SFT + RLVR | 2026-08-24 |
| CHIMERA: Compact Synthetic Data for Generalizable LLM Reasoning | | 9.0 | Qwen3-4B-Thinking-2507 (4B) fine-tuned on Chimera (SFT+RL) | 2026-03-01 |
| Gemma 4 Technical Report | ✓ Link | 8.7 | Gemma 4 26B-A4B | 2026-07-02 |
| LFM2.5-24B-A2B: Built for Fast On-Device Agents | | 7.40 | LFM2.5-24B-A2B | 2026-10-13 |
| Gemma 4 Technical Report | ✓ Link | 5.2 | Gemma 4 12B Unified | 2026-07-02 |
| Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients | | 5.1 | ZPPO (Qwen3.5-2B) | 2026-06-16 |
| Kimi K2: Open Agentic Intelligence | ✓ Link | 4.7 | Kimi K2 | 2025-07-28 |
| Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models | | 2.0 | MELT-1.6B | 2026-05-08 |