Deduction
Subcategory of Reasoning. 161 models scored.
Model Leaderboard
All models ranked by their Deduction subcategory score.
| # | Model | Deduction | Reasoning | Overall |
|---|---|---|---|---|
| 1 | GPT-5.6 Sol (Reasoning) | 98.99% | 96.86% | 95.15% |
| 2 | GPT-5.4 (Reasoning) | 95.95% | 94.89% | 93.85% |
| 3 | Grok 4.5 (Reasoning, High) | 95.35% | 94.45% | 94.12% |
| 4 | Gemini 3.6 Flash (Reasoning) | 93.67% | 94.48% | 94.90% |
| 5 | GPT-5.5 (Reasoning) | 92.17% | 92.87% | 93.72% |
| 6 | Claude Opus 4.6 (Reasoning) | 90.64% | 93.19% | 95.06% |
| 7 | MoonshotAI: Kimi K2.6 | 90.61% | 91.69% | 92.57% |
| 8 | GPT-5.6 Terra (Reasoning) | 89.85% | 91.03% | 92.49% |
| 9 | GPT-5.6 Sol | 89.65% | 91.25% | 90.76% |
| 10 | Claude Opus 4.8 (Reasoning) | 88.50% | 91.76% | 92.33% |
| 11 | MoonshotAI: Kimi K3 (Reasoning, High) | 87.24% | 90.48% | 92.99% |
| 12 | Z.AI GLM 5.1 | 87.24% | 90.64% | 93.74% |
| 13 | GPT-5.5 (Reasoning, Low) | 87.08% | 88.87% | 92.51% |
| 14 | Gemini 3.1 Pro (Preview) | 86.52% | 88.20% | 94.08% |
| 15 | Grok 4.5 (Reasoning, Low) | 86.44% | 89.59% | 90.94% |
| 16 | Gemini 2.5 Pro | 86.38% | 89.21% | 88.44% |
| 17 | Claude Opus 5 (Reasoning, Low) | 86.32% | 88.95% | 92.06% |
| 18 | Claude Opus 4.6 | 86.15% | 89.75% | 92.31% |
| 19 | MoonshotAI: Kimi K3 (Reasoning, Low) | 86.14% | 89.46% | 92.07% |
| 20 | MoonshotAI: Kimi K2.5 | 86.07% | 87.99% | 90.86% |
| 21 | Claude Opus 4.7 | 85.68% | 88.29% | 89.90% |
| 22 | Muse Spark 1.1 (Reasoning, Medium) | 85.67% | 89.03% | 93.81% |
| 23 | Gemini 3.5 Flash (Reasoning) | 85.49% | 88.22% | 93.35% |
| 24 | GPT-5.5 | 85.39% | 87.68% | 89.37% |
| 25 | Claude Opus 4.8 (Reasoning, Low) | 85.19% | 88.83% | 91.89% |
| 26 | GPT-5 | 85.13% | 89.34% | 91.48% |
| 27 | Claude Opus 4.7 (Reasoning) | 85.06% | 88.52% | 92.53% |
| 28 | Claude Sonnet 4.6 (Reasoning) | 85.00% | 89.61% | 93.64% |
| 29 | Aion 3.0 | 84.03% | 86.56% | 88.78% |
| 30 | Qwen3.7 Max | 83.59% | 83.47% | 94.55% |
| 31 | Claude Sonnet 5 (Reasoning) | 83.41% | 87.77% | 90.40% |
| 32 | Claude Sonnet 5 (Reasoning, Low) | 83.40% | 87.78% | 90.16% |
| 33 | Gemini 3 Flash (Preview, Reasoning) | 83.35% | 86.16% | 89.93% |
| 34 | DeepSeek V4 Flash (Reasoning) | 83.08% | 85.45% | 88.06% |
| 35 | GPT-5.4 (Reasoning, Low) | 83.06% | 86.99% | 90.91% |
| 36 | Z.AI GLM 5 Turbo | 83.04% | 86.87% | 93.29% |
| 37 | MiniMax M3 | 83.04% | 86.27% | 90.45% |
| 38 | Claude Opus 5 | 82.80% | 85.64% | 89.97% |
| 39 | GPT-5.4 Mini (Reasoning) | 82.62% | 86.10% | 89.82% |
| 40 | Claude Opus 5 (Reasoning) | 82.56% | 85.32% | 91.15% |
| 41 | GPT-5.6 Luna (Reasoning) | 82.29% | 84.74% | 90.61% |
| 42 | Gemini 3.6 Flash (Reasoning, Minimal) | 81.93% | 83.40% | 86.68% |
| 43 | GPT-5.2 | 81.54% | 85.61% | 89.45% |
| 44 | Thinking Machines Inkling (Reasoning) | 81.28% | 86.03% | 90.78% |
| 45 | Z.AI GLM 5.2 (Reasoning, High) | 80.97% | 85.81% | 93.41% |
| 46 | Muse Spark 1.1 (Reasoning, Minimal) | 80.90% | 85.18% | 91.43% |
| 47 | GPT-5.6 Terra | 80.74% | 84.98% | 88.23% |
| 48 | Aion 3.0 Mini | 80.61% | 75.86% | 83.69% |
| 49 | Qwen3.6 Max Preview | 79.68% | 85.79% | 93.72% |
| 50 | Z.AI GLM 4.7 | 79.52% | 82.43% | 87.67% |
| 51 | GPT-5.1 | 79.33% | 82.57% | 90.73% |
| 52 | Z.AI GLM 5 | 79.21% | 84.09% | 89.60% |
| 53 | GPT-5 Mini | 78.44% | 83.31% | 91.31% |
| 54 | Claude Opus 4.5 | 77.87% | 84.25% | 89.60% |
| 55 | Z.AI GLM 4.6 | 77.79% | 81.01% | 87.64% |
| 56 | o4 Mini High | 77.69% | 82.51% | 88.78% |
| 57 | DeepSeek V4 Pro (Reasoning) | 76.94% | 83.52% | 89.28% |
| 58 | GPT-5.4 Nano (Reasoning) | 76.88% | 77.96% | 80.02% |
| 59 | Qwen 3.5 27B | 76.50% | 79.44% | 90.05% |
| 60 | o4 Mini | 76.48% | 80.39% | 86.56% |
| 61 | Nemotron 3 Super | 75.92% | 77.87% | 81.69% |
| 62 | Qwen 3.5 397B A17B | 75.79% | 81.97% | 91.09% |
| 63 | GPT-5.4 | 75.68% | 80.92% | 84.31% |
| 64 | Qwen 3.5 Plus (2026-02-15) | 75.68% | 81.81% | 86.17% |
| 65 | ByteDance Seed 2.0 Lite | 75.38% | 81.47% | 84.27% |
| 66 | Gemini 3 Flash (Preview) | 75.02% | 80.99% | 85.47% |
| 67 | Aion 2.0 | 74.98% | 78.64% | 86.66% |
| 68 | Xiaomi MIMO v2.5 | 74.69% | 81.80% | 83.95% |
| 69 | Gemini 2.5 Flash | 74.16% | 79.09% | 80.61% |
| 70 | Gemma 4 31B (Reasoning) | 74.02% | 80.14% | 89.64% |
| 71 | ByteDance Seed 1.6 | 74.00% | 79.79% | 89.59% |
| 72 | Qwen 3.5 Plus (2026-04-20) | 73.17% | 80.60% | 89.79% |
| 73 | Xiaomi MIMO v2.5 Pro | 73.13% | 79.73% | 86.05% |
| 74 | Qwen 3.6 27B | 72.95% | 79.84% | 88.33% |
| 75 | Z.AI GLM 4.5 | 72.94% | 77.00% | 84.95% |
| 76 | Gemini 2.5 Flash (Reasoning) | 72.79% | 72.63% | 84.14% |
| 77 | Gemini 3.5 Flash (Reasoning, Minimal) | 72.73% | 78.93% | 85.88% |
| 78 | ByteDance Seed 2.0 Mini | 72.65% | 79.82% | 85.69% |
| 79 | Gemma 3 12B | 72.50% | 56.28% | 76.07% |
| 80 | GPT-4o, Aug. 6th (temp=0) | 72.24% | 75.40% | 82.18% |
| 81 | Claude Sonnet 4.6 | 72.09% | 79.62% | 90.66% |
| 82 | Qwen 3.5 Flash | 71.85% | 79.34% | 85.66% |
| 83 | Gemini 3.5 Flash Lite (Reasoning, Minimal) | 71.77% | 73.22% | 84.45% |
| 84 | MiniMax M2.5 | 71.72% | 74.33% | 86.71% |
| 85 | Gemini 2.5 Flash Lite (Reasoning) | 71.64% | 72.51% | 83.10% |
| 86 | GPT-OSS 120B | 71.54% | 77.00% | 84.81% |
| 87 | Inception Mercury 2 | 71.53% | 76.23% | 81.99% |
| 88 | Qwen 3.6 Flash | 71.49% | 79.51% | 89.31% |
| 89 | Gemini 3.5 Flash Lite (Reasoning) | 71.41% | 74.16% | 84.34% |
| 90 | Qwen 3.5 122B | 71.05% | 79.24% | 90.32% |
| 91 | Claude Sonnet 4.5 | 70.37% | 78.61% | 87.54% |
| 92 | DeepSeek V3.2 | 69.80% | 75.48% | 82.22% |
| 93 | Gemma 4 31B | 69.61% | 78.18% | 85.23% |
| 94 | Mistral Medium 3.1 | 69.60% | 71.84% | 76.08% |
| 95 | Mistral Large 2 | 69.57% | 75.76% | 81.50% |
| 96 | Mistral Large 3 | 69.47% | 75.26% | 84.29% |
| 97 | Gemma 4 26B (Reasoning) | 68.86% | 74.05% | 89.02% |
| 98 | GPT-4.1 | 68.81% | 74.40% | 86.82% |
| 99 | Qwen 3.6 35B | 68.80% | 76.37% | 87.66% |
| 100 | Qwen 3.5 35B | 68.75% | 77.89% | 87.01% |
| 101 | Nemotron 3 Nano | 68.20% | 65.05% | 74.50% |
| 102 | Ministral 3 14B | 68.13% | 67.43% | 70.45% |
| 103 | Grok 4.3 (Reasoning) | 68.02% | 75.64% | 90.99% |
| 104 | Z.AI GLM 4.7 Flash | 67.90% | 69.68% | 82.21% |
| 105 | MiniMax M2.7 | 67.57% | 72.76% | 86.23% |
| 106 | DeepSeek V3.1 | 67.31% | 72.08% | 82.35% |
| 107 | Claude Opus 4 | 67.28% | 76.72% | 87.22% |
| 108 | Claude Sonnet 4 | 67.21% | 75.86% | 87.64% |
| 109 | ByteDance Seed 1.6 Flash | 67.00% | 63.50% | 70.92% |
| 110 | Gemini 3.1 Flash Lite (Reasoning) | 66.95% | 75.32% | 85.91% |
| 111 | Claude Sonnet 5 | 66.73% | 75.35% | 87.34% |
| 112 | DeepSeek V3 (2025-03-24) | 66.68% | 69.32% | 79.93% |
| 113 | GPT-5.6 Luna | 66.67% | 72.88% | 85.06% |
| 114 | Gemini 3.1 Flash Lite | 66.65% | 74.62% | 85.09% |
| 115 | Gemini 3.1 Flash Lite (Preview) | 66.64% | 75.75% | 85.41% |
| 116 | Gemma 4 26B | 66.56% | 73.83% | 84.89% |
| 117 | GPT-4o, Aug. 6th (temp=1) | 66.52% | 70.58% | 81.28% |
| 118 | GPT-5.4 Mini (Reasoning, Low) | 66.36% | 74.50% | 83.57% |
| 119 | Grok 4.20 | 66.13% | 72.46% | 81.21% |
| 120 | GPT-5 Nano | 66.00% | 70.50% | 80.16% |
| 121 | GPT-5.4 Nano (Reasoning, Low) | 65.86% | 64.43% | 77.46% |
| 122 | Qwen 3.5 9B | 65.74% | 70.90% | 84.05% |
| 123 | GPT-5.4 Mini | 64.78% | 70.07% | 80.45% |
| 124 | Qwen 3 32B | 64.44% | 66.35% | 79.37% |
| 125 | Gemini 2.5 Flash Lite | 63.84% | 65.19% | 79.91% |
| 126 | Writer: Palmyra X5 | 63.44% | 69.64% | 78.11% |
| 127 | DeepSeek V3 (2024-12-26) | 63.28% | 69.12% | 82.62% |
| 128 | Laguna XS 2.1 | 62.87% | 67.05% | 75.42% |
| 129 | Mistral Small 4 (Reasoning) | 62.74% | 66.73% | 79.48% |
| 130 | Qwen3 235B A22B Instruct 2507 | 62.64% | 68.43% | 78.07% |
| 131 | DeepSeek-V2 Chat | 62.62% | 70.75% | 84.09% |
| 132 | Laguna S 2.1 | 62.23% | 69.59% | 79.98% |
| 133 | GPT-4o Mini (temp=1) | 62.17% | 62.26% | 77.82% |
| 134 | GPT-4.1 Nano | 61.83% | 54.55% | 69.90% |
| 135 | DeepSeek V4 Pro | 61.62% | 71.72% | 82.05% |
| 136 | Mistral Small 4 | 61.55% | 61.67% | 75.23% |
| 137 | Z.AI GLM 4.5 Air | 60.93% | 68.46% | 80.74% |
| 138 | GPT-4o Mini (temp=0) | 60.87% | 62.17% | 76.86% |
| 139 | Gemma 3 4B | 60.65% | 52.86% | 66.33% |
| 140 | GPT-4.1 Mini | 60.26% | 67.35% | 81.40% |
| 141 | Grok 4.3 | 60.21% | 70.17% | 78.00% |
| 142 | Grok 4.20 (Reasoning) | 60.08% | 74.28% | 90.87% |
| 143 | GPT-5.4 Nano | 59.95% | 58.53% | 72.16% |
| 144 | Hermes 3 405B | 59.50% | 64.59% | 80.80% |
| 145 | Claude Haiku 4.5 | 58.43% | 67.77% | 83.36% |
| 146 | Gemma 3 27B | 58.39% | 61.51% | 75.70% |
| 147 | Ministral 8B | 58.16% | 63.20% | 63.77% |
| 148 | Cydonia 24B V4.1 | 58.13% | 56.95% | 72.68% |
| 149 | Hermes 3 70B | 57.86% | 56.20% | 69.74% |
| 150 | Qwen 2.5 72B | 57.16% | 61.71% | 73.17% |
| 151 | Mistral Small 3.2 24B | 57.06% | 63.87% | 77.36% |
| 152 | WizardLM 2 8x22b | 56.40% | 60.41% | 71.45% |
| 153 | Llama 3.1 70B | 55.78% | 61.45% | 77.41% |
| 154 | Thinking Machines Inkling | 54.93% | 64.85% | 73.96% |
| 155 | Arcee AI: Trinity Mini | 54.53% | 55.45% | 67.68% |
| 156 | DeepSeek V4 Flash | 53.69% | 68.68% | 82.02% |
| 157 | Mistral NeMO | 52.31% | 44.87% | 63.80% |
| 158 | Cohere Command R+ (Aug. 2024) | 51.97% | 51.60% | 67.04% |
| 159 | Ministral 3 8B | 50.65% | 59.53% | 69.98% |
| 160 | Ministral 3 3B | 50.62% | 54.06% | 65.02% |
| 161 | Ministral 3B | 49.47% | 53.21% | 59.25% |