| # | Model | Score | Relative | Setting | Released | Run | Evidence |
|---|---|---|---|---|---|---|---|
| 1 | GPT-6 Astra OpenAI | 98.5% | high | 3 Sep 2026 | — | Source ↗ | |
| 2 | GPT-6 Astra (xhigh) OpenAI | 98.5% | xhigh | 3 Sep 2026 | — | Source ↗ | |
| 3 | Claude Fable 5 Anthropic | 98.5% | max | 9 Jun 2026 | — | Source ↗ | |
| 4 | Claude Fable 5 (xhigh) Anthropic | 98.5% | xhigh | 9 Jun 2026 | — | Source ↗ | |
| 5 | Gemini 3.1 Pro Preview Google | 98.0% | — | 19 Feb 2026 | — | Source ↗ | |
| 6 | GPT-6 Astra (medium) OpenAI | 97.5% | medium | 3 Sep 2026 | — | Source ↗ | |
| 7 | Claude Fable 5.1 Anthropic | 97.5% | max | 1 Sep 2026 | — | Source ↗ | |
| 8 | Claude Opus 5 Anthropic | 97.5% | max | 24 Jul 2026 | — | Source ↗ | |
| 9 | GPT-5.6 Sol (xhigh) OpenAI | 97.5% | xhigh | 9 Jul 2026 | — | Source ↗ | |
| 10 | GPT-5.6 Sol OpenAI | 97.0% | high | 9 Jul 2026 | — | Source ↗ | |
| 11 | GPT-6 Astra (low) OpenAI | 96.5% | low | 3 Sep 2026 | — | Source ↗ | |
| 12 | Claude Fable 5.1 (xhigh) Anthropic | 96.5% | xhigh | 1 Sep 2026 | — | Source ↗ | |
| 13 | GPT-5.6 Terra OpenAI | 96.5% | max | 9 Jul 2026 | — | Source ↗ | |
| 14 | GPT-5.5 Pro OpenAI | 96.5% | high | 23 Apr 2026 | — | Source ↗ | |
| 15 | Gemini 3 Deep Think Google DeepMind | 96.0% | — | 12 Feb 2026 | — | Source ↗ | |
| 16 | Gemini 3.7 Flash Google | 95.5% | high | 13 Aug 2026 | — | Source ↗ | |
| 17 | GPT-5.5 Pro (xhigh) OpenAI | 95.0% | xhigh | 23 Apr 2026 | — | Source ↗ | |
| 18 | GPT-5.5 (xhigh) OpenAI | 95.0% | xhigh | 23 Apr 2026 | — | Source ↗ | |
| 19 | Claude Fable 5.1 (medium) Anthropic | 94.5% | medium | 1 Sep 2026 | — | Source ↗ | |
| 20 | Kimi K3 Open source MoonshotAI | 94.5% | max | 16 Jul 2026 | — | Source ↗ | |
| 21 | GPT-5.5 OpenAI | 94.5% | high | 23 Apr 2026 | — | Source ↗ | |
| 22 | GPT-5.4 Pro (xhigh) OpenAI | 94.5% | xhigh | 5 Mar 2026 | — | Source ↗ | |
| 23 | GPT-5.6 Terra (xhigh) OpenAI | 94.0% | xhigh | 9 Jul 2026 | — | Source ↗ | |
| 24 | Claude Opus 4.6 (120k thinking) Anthropic | 94.0% | 120K | 5 Feb 2026 | — | Source ↗ | |
| 25 | GPT-5.4 (xhigh) OpenAI | 93.7% | xhigh | 5 Mar 2026 | — | Source ↗ | |
| 26 | Claude Opus 4.7 Anthropic | 93.5% | high | 16 Apr 2026 | — | Source ↗ | |
| 27 | GPT-5.4 OpenAI | 92.7% | high | 5 Mar 2026 | — | Source ↗ | |
| 28 | GPT-5.6 Sol (medium) OpenAI | 92.5% | medium | 9 Jul 2026 | — | Source ↗ | |
| 29 | Claude Fable 5 (medium) Anthropic | 92.5% | medium | 9 Jun 2026 | — | Source ↗ | |
| 30 | Claude Opus 4.8 Anthropic | 92.5% | max | 28 May 2026 | — | Source ↗ | |
| 31 | Gemini 3.5 Flash Google | 92.5% | high | 19 May 2026 | — | Source ↗ | |
| 32 | GPT-5.5 (medium) OpenAI | 92.2% | medium | 23 Apr 2026 | — | Source ↗ | |
| 33 | Gemini 3.7 Flash (medium) Google DeepMind | 91.2% | medium | 13 Aug 2026 | — | Source ↗ | |
| 34 | Gemini 3.6 Flash Google | 91.2% | high | 21 Jul 2026 | — | Source ↗ | |
| 35 | Claude Opus 4.7 (low) Anthropic | 91.0% | low | 16 Apr 2026 | — | Source ↗ | |
| 36 | Claude Opus 4.7 (medium) Anthropic | 91.0% | medium | 16 Apr 2026 | — | Source ↗ | |
| 37 | DeepSeek V4 Pro 0813 (low) Open source DeepSeek | 90.5% | low | 13 Aug 2026 | — | Source ↗ | |
| 38 | Claude Fable 5 (low) Anthropic | 90.5% | low | 9 Jun 2026 | — | Source ↗ | |
| 39 | GPT-5.2 Pro OpenAI | 90.5% | xhigh | 11 Dec 2025 | — | Source ↗ | |
| 40 | Claude Fable 5.1 (low) Anthropic | 90.0% | low | 1 Sep 2026 | — | Source ↗ | |
| 41 | DeepSeek V4 Pro 0813 Open source DeepSeek | 90.0% | max | 13 Aug 2026 | — | Source ↗ | |
| 42 | Grok 4.20 SpaceXAI | 89.5% | — | 17 Feb 2026 | — | Source ↗ | |
| 43 | DeepSeek V4 Flash 0731 Open source DeepSeek | 89.0% | max | 31 Jul 2026 | — | Source ↗ | |
| 44 | GPT-5.6 Luna OpenAI | 88.0% | max | 9 Jul 2026 | — | Source ↗ | |
| 45 | GPT-5.6 Luna (xhigh) OpenAI | 87.7% | xhigh | 9 Jul 2026 | — | Source ↗ | |
| 46 | Grok 4.6 (medium) xAI | 87.5% | medium | 12 Aug 2026 | — | Source ↗ | |
| 47 | Grok 4.5 (medium) xAI | 87.2% | medium | 8 Jul 2026 | — | Source ↗ | |
| 48 | Grok 4.6 SpaceXAI | 87.0% | high | 12 Aug 2026 | — | Source ↗ | |
| 49 | Grok 4.6 (xhigh) xAI | 87.0% | xhigh | 12 Aug 2026 | — | Source ↗ | |
| 50 | Claude Sonnet 4.6 Anthropic | 86.5% | high | 17 Feb 2026 | — | Source ↗ | |
| 51 | GPT-5.2 (xhigh) OpenAI | 86.2% | xhigh | 11 Dec 2025 | — | Source ↗ | |
| 52 | GPT-5.4 (medium) OpenAI | 86.2% | medium | 5 Mar 2026 | — | Source ↗ | |
| 53 | GPT-6 Astra (none) OpenAI | 86.0% | none | 3 Sep 2026 | — | Source ↗ | |
| 54 | Grok 4.5 SpaceXAI | 85.7% | high | 8 Jul 2026 | — | Source ↗ | |
| 55 | Gemini 3.7 Flash (low) Google DeepMind | 85.2% | low | 13 Aug 2026 | — | Source ↗ | |
| 56 | Gemini 3 Flash Preview Google | 84.7% | — | 17 Dec 2025 | — | Source ↗ | |
| 57 | DeepSeek V4 Flash 0731 (low) Open source DeepSeek | 84.0% | low | 31 Jul 2026 | — | Source ↗ | |
| 58 | Inkling Small (xhigh) Open source Thinking Machines | 84.0% | xhigh | 15 Jul 2026 | — | Source ↗ | |
| 59 | Gemini 3.6 Flash (medium) Google DeepMind | 83.2% | medium | 21 Jul 2026 | — | Source ↗ | |
| 60 | Claude Opus 4.5 (64k thinking) Anthropic | 80.0% | 64K | 24 Nov 2025 | — | Source ↗ | |
| 61 | Inkling Open source Thinking Machines | 79.5% | — | 15 Jul 2026 | — | Source ↗ | |
| 62 | Grok 4.5 (low) xAI | 79.2% | low | 8 Jul 2026 | — | Source ↗ | |
| 63 | GPT-5.2 OpenAI | 78.7% | high | 11 Dec 2025 | — | Source ↗ | |
| 64 | Inkling Small Open source Thinking Machines | 78.0% | high | 15 Jul 2026 | — | Source ↗ | |
| 65 | GPT-5.6 Terra (medium) OpenAI | 77.0% | medium | 9 Jul 2026 | — | Source ↗ | |
| 66 | GLM 5.2 Open source Z.ai | 77.0% | unknown | 16 Jun 2026 | — | Source ↗ | |
| 67 | GPT-5.5 (low) OpenAI | 76.2% | low | 23 Apr 2026 | — | Source ↗ | |
| 68 | Claude Opus 4.5 (32k thinking) Anthropic | 75.8% | 32K | 24 Nov 2025 | — | Source ↗ | |
| 69 | Gemini 3 Pro Preview Google DeepMind | 75.0% | — | 18 Nov 2025 | — | Source ↗ | |
| 70 | Grok 4.6 (low) xAI | 74.8% | low | 12 Aug 2026 | — | Source ↗ | |
| 71 | GPT-5.6 Sol (low) OpenAI | 74.5% | low | 9 Jul 2026 | — | Source ↗ | |
| 72 | GPT-5.1 OpenAI | 72.8% | high | 13 Nov 2025 | — | Source ↗ | |
| 73 | GPT-5.2 (medium) OpenAI | 72.7% | medium | 11 Dec 2025 | — | Source ↗ | |
| 74 | Claude Opus 4.5 (16k thinking) Anthropic | 72.0% | 16K | 24 Nov 2025 | — | Source ↗ | |
| 75 | GPT-5 Pro OpenAI | 70.2% | high | 7 Oct 2025 | — | Source ↗ | |
| 76 | GPT-5.4 (low) OpenAI | 68.2% | low | 5 Mar 2026 | — | Source ↗ | |
| 77 | Inkling Small (medium) Open source Thinking Machines | 67.0% | medium | 15 Jul 2026 | — | Source ↗ | |
| 78 | Grok 4 xAI | 66.7% | — | 9 Jul 2025 | — | Source ↗ | |
| 79 | GPT-5 OpenAI | 65.7% | high | 7 Aug 2025 | — | Source ↗ | |
| 80 | Kimi K3 (low) Open source Moonshot | 65.7% | low | 16 Jul 2026 | — | Source ↗ | |
| 81 | Kimi K2.5 Open source MoonshotAI | 65.3% | — | 27 Jan 2026 | — | Source ↗ | |
| 82 | GPT-5.4 mini (xhigh) OpenAI | 63.7% | xhigh | 17 Mar 2026 | — | Source ↗ | |
| 83 | MiniMax M2.5 Open source MiniMax | 63.7% | — | 12 Feb 2026 | — | Source ↗ | |
| 84 | Claude Sonnet 4.5 (32k thinking) Anthropic | 63.7% | 32K | 29 Sep 2025 | — | Source ↗ | |
| 85 | o3 OpenAI | 60.8% | high | 16 Apr 2025 | — | Source ↗ | |
| 86 | GPT-5.6 Terra (low) OpenAI | 60.2% | low | 9 Jul 2026 | — | Source ↗ | |
| 87 | o3 Pro OpenAI | 59.3% | high | 10 Jun 2025 | — | Source ↗ | |
| 88 | Claude Opus 4.5 (8k thinking) Anthropic | 58.7% | 8K | 24 Nov 2025 | — | Source ↗ | |
| 89 | o4 Mini OpenAI | 58.7% | high | 16 Apr 2025 | — | Source ↗ | |
| 90 | GPT-5.4 Mini OpenAI | 58.0% | high | 17 Mar 2026 | — | Source ↗ | |
| 91 | GPT-5.1 (medium) OpenAI | 57.7% | medium | 13 Nov 2025 | — | Source ↗ | |
| 92 | DeepSeek-V3.2 (Thinking; Novita) Open source DeepSeek | 57.0% | — | 1 Dec 2025 | — | Source ↗ | |
| 93 | GPT-5.6 Luna (medium) OpenAI | 56.5% | medium | 9 Jul 2026 | — | Source ↗ | |
| 94 | GPT-5 (medium) OpenAI | 56.2% | medium | 7 Aug 2025 | — | Source ↗ | |
| 95 | GPT-5.2 (low) OpenAI | 55.7% | low | 11 Dec 2025 | — | Source ↗ | |
| 96 | GPT-5 Mini OpenAI | 54.3% | high | 7 Aug 2025 | — | Source ↗ | |
| 97 | o3 (medium) OpenAI | 53.8% | medium | 16 Apr 2025 | — | Source ↗ | |
| 98 | Gemini 3.5 Flash Lite Google | 53.5% | high | 21 Jul 2026 | — | Source ↗ | |
| 99 | Inkling Small (low) Open source Thinking Machines | 52.5% | low | 15 Jul 2026 | — | Source ↗ | |
| 100 | GPT-5.4 nano (xhigh) OpenAI | 51.5% | xhigh | 17 Mar 2026 | — | Source ↗ | |
| 101 | Inkling Small (minimal) Open source Thinking Machines | 48.5% | minimal | 15 Jul 2026 | — | Source ↗ | |
| 102 | Grok 4 Fast xAI | 48.5% | — | 19 Sep 2025 | — | Source ↗ | |
| 103 | Claude Sonnet 4.5 (16k thinking) Anthropic | 48.3% | 16K | 29 Sep 2025 | — | Source ↗ | |
| 104 | Claude Haiku 4.5 Anthropic | 47.7% | 32K | 15 Oct 2025 | — | Source ↗ | |
| 105 | Claude Sonnet 4.5 (8k thinking) Anthropic | 46.5% | 8K | 29 Sep 2025 | — | Source ↗ | |
| 106 | GLM 5 Open source Z.ai | 44.7% | — | 11 Feb 2026 | — | Source ↗ | |
| 107 | GPT-5 (low) OpenAI | 44.0% | low | 7 Aug 2025 | — | Source ↗ | |
| 108 | o4-mini (medium) OpenAI | 41.8% | medium | 16 Apr 2025 | — | Source ↗ | |
| 109 | o3 (low) OpenAI | 41.5% | low | 16 Apr 2025 | — | Source ↗ | |
| 110 | Gemini 2.5 Pro (Jun 2025) Google DeepMind | 41.0% | 16K | 5 Jun 2025 | — | Source ↗ | |
| 111 | GPT-5.4 mini (medium) OpenAI | 40.8% | medium | 17 Mar 2026 | — | Source ↗ | |
| 112 | Claude Opus 4.5 (no thinking) Anthropic | 40.0% | — | 24 Nov 2025 | — | Source ↗ | |
| 113 | tiny-recursion-model | 40.0% | — | 6 Oct 2025 | — | Source ↗ | |
| 114 | Claude Sonnet 4 Anthropic | 40.0% | 16K | 22 May 2025 | — | Source ↗ | |
| 115 | GPT-5.4 Nano OpenAI | 38.2% | high | 17 Mar 2026 | — | Source ↗ | |
| 116 | GPT-5 mini (medium) OpenAI | 37.3% | medium | 7 Aug 2025 | — | Source ↗ | |
| 117 | Claude Opus 4 Anthropic | 35.7% | 16K | 22 May 2025 | — | Source ↗ | |
| 118 | o3 Mini OpenAI | 34.5% | high | 31 Jan 2025 | — | Source ↗ | |
| 119 | GPT-5.6 Luna (low) OpenAI | 34.2% | low | 9 Jul 2026 | — | Source ↗ | |
| 120 | Gemini 2.5 Flash (May 2025) Google DeepMind | 33.3% | 16K | 20 May 2025 | — | Source ↗ | |
| 121 | GPT-5.1 (low) OpenAI | 33.2% | low | 13 Nov 2025 | — | Source ↗ | |
| 122 | GPT-5.4 nano (medium) OpenAI | 33.0% | medium | 17 Mar 2026 | — | Source ↗ | |
| 123 | Gemini 2.5 Pro Preview (Mar 2025) Google DeepMind | 33.0% | — | 31 Mar 2025 | — | Source ↗ | |
| 124 | Gemini 3.5 Flash-Lite (medium) Google DeepMind | 32.3% | medium | 21 Jul 2026 | — | Source ↗ | |
| 125 | Gemini 2.5 Flash (Apr 2025) Google DeepMind | 32.3% | — | 17 Apr 2025 | — | Source ↗ | |
| 126 | Gemini 2.5 Pro Preview (Jun 2025) Google DeepMind | 31.3% | 1K | 5 Jun 2025 | — | Source ↗ | |
| 127 | Claude Sonnet 4.5 (1k thinking) Anthropic | 31.0% | 1K | 29 Sep 2025 | — | Source ↗ | |
| 128 | o1 (medium) OpenAI | 30.7% | medium | 17 Dec 2024 | — | Source ↗ | |
| 129 | Claude 3.7 Sonnet (16k thinking) Anthropic | 28.6% | 16K | 24 Feb 2025 | — | Source ↗ | |
| 130 | codex-mini-2025-05-16 | 27.3% | — | 16 May 2025 | — | Source ↗ | |
| 131 | o1 (low) OpenAI | 27.2% | low | 17 Dec 2024 | — | Source ↗ | |
| 132 | GPT-5 mini (low) OpenAI | 26.3% | low | 7 Aug 2025 | — | Source ↗ | |
| 133 | Claude Sonnet 4.5 (no thinking) Anthropic | 25.5% | — | 29 Sep 2025 | — | Source ↗ | |
| 134 | o1 Pro (low) OpenAI | 23.3% | low | 19 Mar 2025 | — | Source ↗ | |
| 135 | o3-mini (medium) OpenAI | 22.3% | medium | 31 Jan 2025 | — | Source ↗ | |
| 136 | o4-mini (low) OpenAI | 21.3% | low | 16 Apr 2025 | — | Source ↗ | |
| 137 | DeepSeek-R1 (May 2025) Open source DeepSeek | 21.2% | — | 28 May 2025 | — | Source ↗ | |
| 138 | Claude 3.7 Sonnet (8k thinking) Anthropic | 21.2% | 8K | 24 Feb 2025 | — | Source ↗ | |
| 139 | GPT-5 nano (medium) OpenAI | 20.7% | medium | 7 Aug 2025 | — | Source ↗ | |
| 140 | GPT-5.4 nano (low) OpenAI | 18.3% | low | 17 Mar 2026 | — | Source ↗ | |
| 141 | o1-preview OpenAI | 18.0% | — | 12 Sep 2024 | — | Source ↗ | |
| 142 | GPT-5 Nano OpenAI | 16.7% | high | 7 Aug 2025 | — | Source ↗ | |
| 143 | Grok-3 mini xAI | 16.5% | low | 24 Jun 2025 | — | Source ↗ | |
| 144 | R1 Open source DeepSeek | 15.8% | — | 20 Jan 2025 | — | Source ↗ | |
| 145 | o1-mini (medium) OpenAI | 14.0% | medium | 12 Sep 2024 | — | Source ↗ | |
| 146 | o1-mini (unknown thinking) OpenAI | 14.0% | unknown | 12 Sep 2024 | — | Source ↗ | |
| 147 | Claude 3.7 Sonnet Anthropic | 13.6% | — | 24 Feb 2025 | — | Source ↗ | |
| 148 | DeepSeek V4 Pro 0813 (none) Open source DeepSeek | 13.0% | none | 13 Aug 2026 | — | Source ↗ | |
| 149 | GPT-5.4 mini (low) OpenAI | 13.0% | low | 17 Mar 2026 | — | Source ↗ | |
| 150 | GPT-5.2 (none) OpenAI | 12.3% | none | 11 Dec 2025 | — | Source ↗ | |
| 151 | GPT-5.2 (unknown thinking) OpenAI | 12.3% | unknown | 11 Dec 2025 | — | Source ↗ | |
| 152 | DeepSeek V4 Flash 0731 (none) Open source DeepSeek | 11.8% | none | 31 Jul 2026 | — | Source ↗ | |
| 153 | Claude 3.7 Sonnet (1k thinking) Anthropic | 11.6% | 1K | 24 Feb 2025 | — | Source ↗ | |
| 154 | Qwen3-235B-A22B-Instruct (Jul 2025) Open source Alibaba | 11.0% | — | 25 Jul 2025 | — | Source ↗ | |
| 155 | GPT-4.5 Preview (Feb 2025) OpenAI | 10.3% | — | 27 Feb 2025 | — | Source ↗ | |
| 156 | Magistral Medium 1.1 Mistral AI | 6.1% | — | 10 Jun 2025 | — | Source ↗ | |
| 157 | GPT-5 (minimal) OpenAI | 6.0% | minimal | 7 Aug 2025 | — | Source ↗ | |
| 158 | GPT-5.1 (no thinking) OpenAI | 5.8% | none | 13 Nov 2025 | — | Source ↗ | |
| 159 | Inkling Small (none) Open source Thinking Machines | 5.5% | none | 15 Jul 2026 | — | Source ↗ | |
| 160 | GPT-4.1 OpenAI | 5.5% | — | 14 Apr 2025 | — | Source ↗ | |
| 161 | Grok 3 xAI | 5.5% | — | 9 Apr 2025 | — | Source ↗ | |
| 162 | GPT-5 mini (minimal) OpenAI | 5.3% | minimal | 7 Aug 2025 | — | Source ↗ | |
| 163 | Magistral Small 1.0 Open source Mistral AI | 5.0% | — | 10 Jun 2025 | — | Source ↗ | |
| 164 | GPT-4o (Nov 2024) OpenAI | 4.5% | — | 20 Nov 2024 | — | Source ↗ | |
| 165 | Llama 4 Maverick Open source Meta | 4.4% | — | 6 Apr 2025 | — | Source ↗ | |
| 166 | GPT-5 nano (low) OpenAI | 4.0% | low | 7 Aug 2025 | — | Source ↗ | |
| 167 | GPT-4.1 Mini OpenAI | 3.5% | — | 14 Apr 2025 | — | Source ↗ | |
| 168 | GPT-5 nano (minimal) OpenAI | 1.5% | minimal | 7 Aug 2025 | — | Source ↗ | |
| 169 | Llama 4 Scout Open source Meta | 0.5% | — | 5 Apr 2025 | — | Source ↗ | |
| 170 | GPT-4.1 Nano OpenAI | 0.0% | — | 14 Apr 2025 | — | Source ↗ |
Caveats: settings (reasoning effort, agent scaffold) differ between rows and materially affect scores; the best reported setting per model is shown. Source: Epoch AI, CC BY 4.0.