Rule Following
12 scenarios across 1 subcategory. 146 models scored.
Subcategories
| Subcategory | Avg Score | Best Model | Best Score |
|---|---|---|---|
| Constraint Adherence | 63.71% | Qwen3.7 Max | 95.76% |
Model Leaderboard
All models ranked by their Rule Following category score.
| # | Model | Rule Following | Constraint Adherence | Overall |
|---|---|---|---|---|
| 1 | Qwen3.7 Max | 95.76% | 95.76% | 94.55% |
| 2 | Gemini 3.5 Flash (Reasoning) | 92.04% | 92.04% | 93.35% |
| 3 | Gemini 3.1 Pro (Preview) | 91.21% | 91.21% | 94.08% |
| 4 | Claude Opus 4.6 (Reasoning) | 89.78% | 89.78% | 95.06% |
| 5 | Z.AI GLM 5.1 | 88.41% | 88.41% | 93.74% |
| 6 | Grok 4.5 (Reasoning, High) | 86.79% | 86.79% | 94.12% |
| 7 | Z.AI GLM 5 Turbo | 86.78% | 86.78% | 93.29% |
| 8 | GPT-5.6 Sol (Reasoning) | 85.88% | 85.88% | 95.15% |
| 9 | Claude Sonnet 4.6 (Reasoning) | 85.73% | 85.73% | 93.64% |
| 10 | Z.AI GLM 5.2 (Reasoning, High) | 85.18% | 85.18% | 93.41% |
| 11 | Gemma 4 31B (Reasoning) | 85.00% | 85.00% | 89.64% |
| 12 | Claude Opus 4.6 | 83.11% | 83.11% | 92.31% |
| 13 | Grok 4.3 (Reasoning) | 82.80% | 82.80% | 90.99% |
| 14 | Qwen3.6 Max Preview | 82.79% | 82.79% | 93.72% |
| 15 | Claude Sonnet 4.6 | 82.50% | 82.50% | 90.66% |
| 16 | Grok 4.20 (Reasoning) | 82.04% | 82.04% | 90.87% |
| 17 | Claude Sonnet 4 | 81.52% | 81.52% | 87.64% |
| 18 | Qwen 3.5 122B | 80.00% | 80.00% | 90.32% |
| 19 | GPT-5.5 (Reasoning) | 79.40% | 79.40% | 93.72% |
| 20 | Qwen 3.5 397B A17B | 79.39% | 79.39% | 91.09% |
| 21 | GPT-5.4 (Reasoning) | 79.29% | 79.29% | 93.85% |
| 22 | ByteDance Seed 1.6 | 77.71% | 77.71% | 89.59% |
| 23 | Qwen 3.6 35B | 77.34% | 77.34% | 87.66% |
| 24 | GPT-5 | 77.13% | 77.13% | 91.48% |
| 25 | MoonshotAI: Kimi K2.6 | 76.94% | 76.94% | 92.57% |
| 26 | GPT-5.5 (Reasoning, Low) | 76.90% | 76.90% | 92.51% |
| 27 | Claude Sonnet 4.5 | 76.80% | 76.80% | 87.54% |
| 28 | Grok 4.5 (Reasoning, Low) | 76.44% | 76.44% | 90.94% |
| 29 | GPT-5 Mini | 76.44% | 76.44% | 91.31% |
| 30 | Qwen 3.5 27B | 76.04% | 76.04% | 90.05% |
| 31 | Gemma 4 26B (Reasoning) | 74.75% | 74.75% | 89.02% |
| 32 | Gemini 3 Flash (Preview, Reasoning) | 74.48% | 74.48% | 89.93% |
| 33 | GPT-5.6 Sol | 74.47% | 74.47% | 90.76% |
| 34 | Claude Sonnet 5 | 74.36% | 74.36% | 87.34% |
| 35 | GPT-4o, Aug. 6th (temp=0) | 74.19% | 74.19% | 82.18% |
| 36 | GPT-5.1 | 74.05% | 74.05% | 90.73% |
| 37 | Claude Opus 4.7 (Reasoning) | 74.04% | 74.04% | 92.53% |
| 38 | GPT-5.6 Terra (Reasoning) | 72.99% | 72.99% | 92.49% |
| 39 | GPT-5.6 Terra | 72.82% | 72.82% | 88.23% |
| 40 | DeepSeek V4 Pro (Reasoning) | 72.74% | 72.74% | 89.28% |
| 41 | Gemma 4 31B | 72.72% | 72.72% | 85.23% |
| 42 | o4 Mini High | 72.70% | 72.70% | 88.78% |
| 43 | Claude Opus 4.5 | 72.61% | 72.61% | 89.60% |
| 44 | GPT-5.5 | 72.34% | 72.34% | 89.37% |
| 45 | MoonshotAI: Kimi K2.5 | 72.03% | 72.03% | 90.86% |
| 46 | Gemma 4 26B | 71.75% | 71.75% | 84.89% |
| 47 | Qwen 3.6 Flash | 71.50% | 71.50% | 89.31% |
| 48 | Qwen 3.6 27B | 71.37% | 71.37% | 88.33% |
| 49 | Claude Opus 4.8 (Reasoning, Low) | 70.56% | 70.56% | 91.89% |
| 50 | Claude Opus 4 | 70.37% | 70.37% | 87.22% |
| 51 | Claude Haiku 4.5 | 70.35% | 70.35% | 83.36% |
| 52 | Claude Opus 4.8 (Reasoning) | 70.27% | 70.27% | 92.33% |
| 53 | GPT-5.4 (Reasoning, Low) | 70.02% | 70.02% | 90.91% |
| 54 | MiniMax M3 | 69.18% | 69.18% | 90.45% |
| 55 | Z.AI GLM 4.7 | 69.16% | 69.16% | 87.67% |
| 56 | MiniMax M2.7 | 68.90% | 68.90% | 86.23% |
| 57 | Aion 3.0 | 68.80% | 68.80% | 88.78% |
| 58 | DeepSeek-V2 Chat | 68.78% | 68.78% | 84.09% |
| 59 | Claude Sonnet 5 (Reasoning) | 68.34% | 68.34% | 90.40% |
| 60 | Claude Opus 4.7 | 68.08% | 68.08% | 89.90% |
| 61 | DeepSeek V3 (2025-03-24) | 67.94% | 67.94% | 79.93% |
| 62 | GPT-4o, Aug. 6th (temp=1) | 67.91% | 67.91% | 81.28% |
| 63 | Z.AI GLM 5 | 67.78% | 67.78% | 89.60% |
| 64 | Qwen 3.5 Plus (2026-04-20) | 67.53% | 67.53% | 89.79% |
| 65 | Qwen 3.5 35B | 67.42% | 67.42% | 87.01% |
| 66 | Writer: Palmyra X5 | 67.19% | 67.19% | 78.11% |
| 67 | GPT-5.2 | 67.10% | 67.10% | 89.45% |
| 68 | Gemini 2.5 Flash Lite (Reasoning) | 66.81% | 66.81% | 83.10% |
| 69 | GPT-4.1 | 66.78% | 66.78% | 86.82% |
| 70 | Claude Sonnet 5 (Reasoning, Low) | 66.62% | 66.62% | 90.16% |
| 71 | DeepSeek V3 (2024-12-26) | 66.39% | 66.39% | 82.62% |
| 72 | DeepSeek V3.1 | 66.15% | 66.15% | 82.35% |
| 73 | Z.AI GLM 4.6 | 65.85% | 65.85% | 87.64% |
| 74 | Z.AI GLM 4.7 Flash | 65.63% | 65.63% | 82.21% |
| 75 | Qwen3 235B A22B Instruct 2507 | 65.42% | 65.42% | 78.07% |
| 76 | Gemini 3 Flash (Preview) | 65.14% | 65.14% | 85.47% |
| 77 | GPT-5.6 Luna (Reasoning) | 64.83% | 64.83% | 90.61% |
| 78 | o4 Mini | 64.61% | 64.61% | 86.56% |
| 79 | DeepSeek V4 Flash (Reasoning) | 64.50% | 64.50% | 88.06% |
| 80 | Mistral Large 3 | 64.41% | 64.41% | 84.29% |
| 81 | Xiaomi MIMO v2.5 Pro | 64.29% | 64.29% | 86.05% |
| 82 | Qwen 3.5 Plus (2026-02-15) | 64.21% | 64.21% | 86.17% |
| 83 | Mistral Small 3.2 24B | 64.08% | 64.08% | 77.36% |
| 84 | Z.AI GLM 4.5 | 63.79% | 63.79% | 84.95% |
| 85 | Aion 2.0 | 63.77% | 63.77% | 86.66% |
| 86 | DeepSeek V4 Pro | 63.74% | 63.74% | 82.05% |
| 87 | Llama 3.1 70B | 63.45% | 63.45% | 77.41% |
| 88 | Qwen 3.5 Flash | 63.19% | 63.19% | 85.66% |
| 89 | Mistral Large 2 | 63.05% | 63.05% | 81.50% |
| 90 | MiniMax M2.5 | 62.69% | 62.69% | 86.71% |
| 91 | Gemini 3.1 Flash Lite (Reasoning) | 62.26% | 62.26% | 85.91% |
| 92 | Mistral Small 4 | 62.17% | 62.17% | 75.23% |
| 93 | Gemini 3.1 Flash Lite | 61.21% | 61.21% | 85.09% |
| 94 | Gemma 3 12B | 61.05% | 61.05% | 76.07% |
| 95 | Aion 3.0 Mini | 60.98% | 60.98% | 83.69% |
| 96 | Qwen 3.5 9B | 60.98% | 60.98% | 84.05% |
| 97 | Gemini 2.5 Pro | 60.89% | 60.89% | 88.44% |
| 98 | Xiaomi MIMO v2.5 | 60.75% | 60.75% | 83.95% |
| 99 | Gemini 3.5 Flash (Reasoning, Minimal) | 60.38% | 60.38% | 85.88% |
| 100 | Mistral Small 4 (Reasoning) | 60.28% | 60.28% | 79.48% |
| 101 | Gemini 2.5 Flash (Reasoning) | 59.97% | 59.97% | 84.14% |
| 102 | Gemini 2.5 Flash Lite | 59.96% | 59.96% | 79.91% |
| 103 | Grok 4.20 | 59.71% | 59.71% | 81.21% |
| 104 | Hermes 3 405B | 59.17% | 59.17% | 80.80% |
| 105 | Gemini 3.1 Flash Lite (Preview) | 59.04% | 59.04% | 85.41% |
| 106 | GPT-4o Mini (temp=0) | 58.84% | 58.84% | 76.86% |
| 107 | ByteDance Seed 2.0 Mini | 58.77% | 58.77% | 85.69% |
| 108 | Cohere Command R+ (Aug. 2024) | 58.70% | 58.70% | 67.04% |
| 109 | GPT-4.1 Mini | 58.59% | 58.59% | 81.40% |
| 110 | GPT-5.4 | 58.11% | 58.11% | 84.31% |
| 111 | GPT-5 Nano | 57.57% | 57.57% | 80.16% |
| 112 | Gemini 2.5 Flash | 57.47% | 57.47% | 80.61% |
| 113 | Nemotron 3 Super | 57.43% | 57.43% | 81.69% |
| 114 | GPT-5.4 Mini (Reasoning) | 57.38% | 57.38% | 89.82% |
| 115 | DeepSeek V4 Flash | 57.32% | 57.32% | 82.02% |
| 116 | GPT-4o Mini (temp=1) | 56.50% | 56.50% | 77.82% |
| 117 | GPT-OSS 120B | 55.03% | 55.03% | 84.81% |
| 118 | GPT-5.6 Luna | 54.70% | 54.70% | 85.06% |
| 119 | Inception Mercury 2 | 54.41% | 54.41% | 81.99% |
| 120 | DeepSeek V3.2 | 53.75% | 53.75% | 82.22% |
| 121 | Hermes 3 70B | 53.00% | 53.00% | 69.74% |
| 122 | Ministral 3 14B | 50.83% | 50.83% | 70.45% |
| 123 | Cydonia 24B V4.1 | 50.36% | 50.36% | 72.68% |
| 124 | Grok 4.3 | 49.02% | 49.02% | 78.00% |
| 125 | Mistral Medium 3.1 | 48.60% | 48.60% | 76.08% |
| 126 | Gemma 3 27B | 47.98% | 47.98% | 75.70% |
| 127 | ByteDance Seed 1.6 Flash | 47.15% | 47.15% | 70.92% |
| 128 | Qwen 3 32B | 46.83% | 46.83% | 79.37% |
| 129 | GPT-5.4 Mini | 46.32% | 46.32% | 80.45% |
| 130 | Z.AI GLM 4.5 Air | 44.11% | 44.11% | 80.74% |
| 131 | Nemotron 3 Nano | 43.47% | 43.47% | 74.50% |
| 132 | GPT-4.1 Nano | 40.88% | 40.88% | 69.90% |
| 133 | ByteDance Seed 2.0 Lite | 36.85% | 36.85% | 84.27% |
| 134 | Mistral NeMO | 34.11% | 34.11% | 63.80% |
| 135 | GPT-5.4 Mini (Reasoning, Low) | 33.99% | 33.99% | 83.57% |
| 136 | GPT-5.4 Nano (Reasoning, Low) | 31.65% | 31.65% | 77.46% |
| 137 | Qwen 2.5 72B | 31.55% | 31.55% | 73.17% |
| 138 | Ministral 3 8B | 31.34% | 31.34% | 69.98% |
| 139 | WizardLM 2 8x22b | 28.27% | 28.27% | 71.45% |
| 140 | GPT-5.4 Nano (Reasoning) | 27.15% | 27.15% | 80.02% |
| 141 | Gemma 3 4B | 26.37% | 26.37% | 66.33% |
| 142 | Ministral 3B | 24.45% | 24.45% | 59.25% |
| 143 | Arcee AI: Trinity Mini | 23.57% | 23.57% | 67.68% |
| 144 | GPT-5.4 Nano | 20.94% | 20.94% | 72.16% |
| 145 | Ministral 3 3B | 15.87% | 15.87% | 65.02% |
| 146 | Ministral 8B | 15.27% | 15.27% | 63.77% |