False Positives
Subcategory of Hallucination. 202 models scored.
Model Leaderboard
All models ranked by their False Positives subcategory score.
| # | Model | False Positives | Hallucination | Overall |
|---|---|---|---|---|
| 1 | Claude Haiku 5.5 (Adaptive) | 99.06% | 96.12% | 89.37% |
| 2 | Muse Spark 1.1 (Reasoning, Minimal) | 98.59% | 95.34% | 91.43% |
| 3 | o4 Mini High | 98.59% | 99.53% | 88.78% |
| 4 | Claude Haiku 5.5 (Reasoning, Low) | 98.59% | 89.61% | 84.75% |
| 5 | Z.AI GLM 5.3 Flash (Reasoning, Low) | 98.13% | 99.36% | 88.76% |
| 6 | Inception Mercury 2 | 98.13% | 93.22% | 81.99% |
| 7 | Z.AI GLM 5 Turbo | 97.97% | 99.32% | 93.29% |
| 8 | Claude Opus 5.5 (Reasoning) | 97.66% | 97.63% | 96.43% |
| 9 | Muse Spark 1.1 (Reasoning, Medium) | 97.50% | 96.79% | 93.81% |
| 10 | Claude Haiku 5.5 (Reasoning, Medium) | 97.41% | 91.20% | 84.17% |
| 11 | Muse Spark 1.2 (Reasoning, Medium) | 97.37% | 97.93% | 94.71% |
| 12 | Claude Opus 4.6 (Reasoning) | 97.19% | 99.06% | 95.06% |
| 13 | Grok 4.5 (Reasoning, High) | 97.19% | 99.06% | 94.12% |
| 14 | Grok 4.5 (Reasoning, Low) | 97.03% | 99.01% | 90.94% |
| 15 | Z.AI GLM 5.2 (Reasoning, High) | 96.98% | 98.99% | 93.41% |
| 16 | Z.AI GLM 5.1 | 96.51% | 98.84% | 93.74% |
| 17 | Gemma 4 26B (Reasoning) | 96.41% | 98.79% | 89.02% |
| 18 | GPT-5.4 Mini (Reasoning, Low) | 96.36% | 98.78% | 83.57% |
| 19 | Gemini 4 Argon (Reasoning, Medium) | 96.25% | 98.69% | 93.81% |
| 20 | Qwen 3.5 Plus (2026-04-20) | 96.25% | 98.38% | 89.79% |
| 21 | Claude Opus 4.7 (Reasoning) | 96.09% | 98.69% | 92.53% |
| 22 | GPT-5.6 Luna (Reasoning, Medium) | 95.94% | 98.65% | 90.61% |
| 23 | Gemini 2.5 Flash Lite (Reasoning) | 95.78% | 96.99% | 83.10% |
| 24 | Muse Spark 1.3 (Reasoning, Medium) | 95.73% | 98.57% | 94.06% |
| 25 | o4 Mini | 95.63% | 98.54% | 86.56% |
| 26 | GPT-5 Mini | 94.85% | 98.28% | 91.31% |
| 27 | Grok 4.6 (Reasoning, High) | 94.84% | 98.25% | 92.76% |
| 28 | GPT-5.4 Mini (Reasoning) | 94.32% | 98.11% | 89.82% |
| 29 | DeepSeek V4 Flash (Reasoning) | 94.32% | 96.92% | 88.06% |
| 30 | Z.AI GLM 5 | 94.11% | 98.04% | 89.60% |
| 31 | Gemini 3.7 Flash (Reasoning, Medium) | 93.91% | 97.97% | 94.85% |
| 32 | Grok 4.3 (Reasoning) | 93.59% | 97.86% | 90.99% |
| 33 | MiniMax M3 | 93.49% | 97.83% | 90.45% |
| 34 | Xiaomi MIMO v2.6 Flash | 93.44% | 97.74% | 87.31% |
| 35 | Gemini 2.5 Flash (Reasoning) | 93.44% | 97.79% | 84.14% |
| 36 | Kimi K3 (Reasoning, Low) | 93.28% | 97.73% | 92.07% |
| 37 | GPT-OSS 120B | 92.97% | 97.63% | 84.81% |
| 38 | Xiaomi MIMO v2.5 | 92.92% | 97.61% | 83.95% |
| 39 | ByteDance Seed 1.6 | 92.71% | 97.57% | 89.59% |
| 40 | GPT-5.2 | 92.64% | 97.55% | 89.45% |
| 41 | Qwen 3.6 27B | 92.33% | 97.42% | 88.33% |
| 42 | Claude Opus 4.8 (Reasoning) | 92.24% | 97.41% | 92.33% |
| 43 | Muse Glimmer 30B (Reasoning, Medium) | 92.14% | 96.19% | 88.81% |
| 44 | Gemini 3.6 Flash (Reasoning) | 92.03% | 97.34% | 94.90% |
| 45 | Claude Opus 5 (Reasoning, Low) | 92.03% | 97.34% | 92.06% |
| 46 | Nemotron 3 Super | 92.03% | 97.34% | 81.69% |
| 47 | Claude Opus 4.8 (Reasoning, Low) | 91.98% | 97.33% | 91.89% |
| 48 | Qwen 3.7 Max | 91.51% | 97.15% | 94.55% |
| 49 | Xiaomi MIMO v2.5 Pro | 91.05% | 97.02% | 86.05% |
| 50 | GPT-5.4 Nano (Reasoning) | 90.95% | 96.98% | 80.02% |
| 51 | Claude Sonnet 4.6 (Reasoning) | 90.94% | 96.98% | 93.64% |
| 52 | Kimi K3 (Reasoning, High) | 90.91% | 96.97% | 92.99% |
| 53 | GPT-5 Nano | 90.90% | 92.99% | 80.16% |
| 54 | Aion 3.5 Mini (Reasoning, High) | 90.78% | 96.91% | 92.36% |
| 55 | Gemini 3.8 Flash (Reasoning, Medium) | 90.63% | 96.88% | 95.72% |
| 56 | Mistral Large 4.0 (Reasoning) | 90.60% | 96.85% | 92.14% |
| 57 | Claude Opus 5 (Reasoning) | 90.57% | 96.86% | 91.15% |
| 58 | Claude Opus 4.6 | 90.40% | 96.80% | 92.31% |
| 59 | Claude Sonnet 5 (Reasoning) | 90.31% | 96.77% | 90.40% |
| 60 | DeepSeek V4 Flash 0731 (Reasoning, High) | 90.10% | 94.32% | 87.97% |
| 61 | Ember-1 (Reasoning, Max) | 90.00% | 96.67% | 95.52% |
| 62 | Gemini 3.6 Flash (Reasoning, Minimal) | 89.59% | 96.53% | 86.68% |
| 63 | GPT-6 Luna (Reasoning, Medium) | 89.58% | 96.53% | 92.40% |
| 64 | Claude Sonnet 5 (Reasoning, Low) | 89.32% | 96.44% | 90.16% |
| 65 | Gemma 4 31B (Reasoning) | 89.11% | 96.37% | 89.64% |
| 66 | Thinking Machines Inkling (Reasoning) | 89.01% | 96.34% | 90.78% |
| 67 | Grok 4.20 (Reasoning) | 88.89% | 96.30% | 90.87% |
| 68 | GPT-6 Luna (Reasoning, High) | 88.70% | 96.23% | 93.95% |
| 69 | GPT-5.1 | 88.65% | 96.22% | 90.73% |
| 70 | Aion 3.5 (Reasoning, High) | 88.39% | 96.13% | 91.90% |
| 71 | Qwen 3.6 Flash | 88.39% | 96.13% | 89.31% |
| 72 | GPT-6.1 Sol (Reasoning, Medium) | 88.28% | 94.90% | 95.88% |
| 73 | Claude Sonnet 5.5 (Reasoning) | 88.28% | 96.09% | 92.77% |
| 74 | Aion 3.0 Mini | 88.15% | 94.08% | 83.69% |
| 75 | GPT-5 | 87.76% | 95.92% | 91.48% |
| 76 | GPT-5.6 Sol | 87.60% | 95.87% | 90.76% |
| 77 | Qwen 3.6 Max Preview | 87.58% | 95.86% | 93.72% |
| 78 | GPT-5.4 (Reasoning, Low) | 87.14% | 95.71% | 90.91% |
| 79 | GPT-5.4 Nano (Reasoning, Low) | 87.13% | 95.71% | 77.46% |
| 80 | DeepSeek V4 Pro (Reasoning) | 86.88% | 95.25% | 89.28% |
| 81 | Gemini 3.5 Flash (Reasoning) | 86.82% | 95.61% | 93.35% |
| 82 | Aion 3.0 | 86.82% | 95.57% | 88.78% |
| 83 | GPT-4.1 | 86.82% | 95.60% | 86.82% |
| 84 | Grok 4.7 (Reasoning, High) | 86.43% | 95.48% | 92.62% |
| 85 | Aion 2.0 | 86.41% | 93.10% | 86.66% |
| 86 | Z.AI GLM 5.3 (Reasoning, Max) | 85.91% | 95.30% | 95.20% |
| 87 | DeepSeek V4.1 Flash (Reasoning, High) | 85.81% | 95.27% | 94.30% |
| 88 | Nemotron 3 Nano | 85.80% | 88.55% | 74.50% |
| 89 | GPT-5.4 (Reasoning) | 85.65% | 95.22% | 93.85% |
| 90 | Z.AI GLM 5.3 Flash (Reasoning, Max) | 85.36% | 95.12% | 94.28% |
| 91 | Gemma 4 31B | 85.26% | 95.09% | 85.23% |
| 92 | Claude Sonnet 4.6 | 84.99% | 94.99% | 90.66% |
| 93 | GPT-5.6 Terra (Reasoning, Medium) | 84.95% | 94.98% | 92.49% |
| 94 | MiniMax M2.5 | 84.66% | 94.72% | 86.71% |
| 95 | Z.AI GLM 4.6 | 84.38% | 94.42% | 87.64% |
| 96 | MiniMax M2.7 | 84.34% | 94.69% | 86.23% |
| 97 | Gemini 3.1 Pro (Preview) | 83.59% | 94.53% | 94.08% |
| 98 | Laguna XS 2.1 | 83.48% | 92.09% | 75.42% |
| 99 | Qwen 3.6 35B | 83.33% | 94.07% | 87.66% |
| 100 | Kimi K2.6 | 83.28% | 94.43% | 92.57% |
| 101 | Qwen 3.7 Flash (Reasoning) | 83.28% | 94.02% | 88.56% |
| 102 | Thinking Machines Inkling Small (Reasoning, High) | 83.20% | 94.39% | 87.39% |
| 103 | Z.AI GLM 4.5 Air | 83.12% | 94.00% | 80.74% |
| 104 | Mistral Small 4 (Reasoning) | 83.11% | 93.18% | 79.48% |
| 105 | ByteDance Seed 1.6 Flash | 83.03% | 90.83% | 70.92% |
| 106 | Z.AI GLM 4.7 | 82.73% | 94.23% | 87.67% |
| 107 | Gemma 4 26B | 82.67% | 94.22% | 84.89% |
| 108 | Hy4 Preview (Reasoning, High) | 82.50% | 94.17% | 94.80% |
| 109 | ByteDance Seed 2.0 Mini | 82.30% | 94.07% | 85.69% |
| 110 | Ministral 3 8B | 82.06% | 94.02% | 69.98% |
| 111 | Kimi K2.5 | 82.03% | 94.01% | 90.86% |
| 112 | Muse Wonder (Reasoning, Medium) | 81.77% | 93.92% | 95.47% |
| 113 | GPT-6 Sol (Reasoning, Medium) | 81.72% | 93.91% | 92.63% |
| 114 | Claude Opus 5 | 81.72% | 93.91% | 89.97% |
| 115 | Xiaomi MIMO v2.6 Pro | 81.51% | 93.78% | 91.50% |
| 116 | Qwen 3.5 9B | 81.32% | 91.81% | 84.05% |
| 117 | GPT-5.6 Luna | 80.10% | 93.37% | 85.06% |
| 118 | Qwen 3.5 122B | 79.87% | 93.29% | 90.32% |
| 119 | Qwen 3.5 27B | 79.86% | 93.29% | 90.05% |
| 120 | Laguna S 2.1 | 79.51% | 93.17% | 79.99% |
| 121 | Gemini 2.5 Pro | 79.16% | 93.05% | 88.44% |
| 122 | Claude Opus 4.7 | 78.84% | 92.95% | 89.90% |
| 123 | Qwen 3.8 Max (Reasoning, XHigh) | 78.28% | 92.76% | 96.21% |
| 124 | Thinking Machines Inkling | 78.10% | 92.70% | 73.96% |
| 125 | Gemini 3 Flash (Preview, Reasoning) | 77.99% | 92.65% | 89.93% |
| 126 | GPT-6 Luna | 77.71% | 92.57% | 83.73% |
| 127 | Z.AI GLM 4.7 Flash | 77.13% | 90.00% | 82.21% |
| 128 | Z.AI GLM 4.5 | 76.65% | 91.83% | 84.95% |
| 129 | GPT-5.5 (Reasoning, Medium) | 76.35% | 92.12% | 93.72% |
| 130 | Claude Haiku 4.5 | 75.80% | 91.93% | 83.36% |
| 131 | Ministral 8B | 75.68% | 91.89% | 63.77% |
| 132 | Qwen 3.8 27B (Reasoning, XHigh) | 75.31% | 90.29% | 92.91% |
| 133 | Qwen 3.8 Flash (Reasoning) | 75.27% | 91.39% | 94.24% |
| 134 | GPT-5.6 Sol (Reasoning, Medium) | 75.16% | 91.72% | 95.15% |
| 135 | Qwen 3.5 35B | 74.87% | 89.24% | 87.01% |
| 136 | DeepSeek V4 Pro 0813 (Reasoning, High) | 74.61% | 88.77% | 90.21% |
| 137 | GPT-5.5 (Reasoning, Low) | 74.11% | 91.37% | 92.51% |
| 138 | Qwen 3.5 Flash | 73.21% | 88.70% | 85.66% |
| 139 | Claude Opus 4.5 | 73.09% | 91.03% | 89.60% |
| 140 | GPT-5.6 Terra | 71.90% | 90.63% | 88.23% |
| 141 | Mistral Large 4.0 | 71.55% | 90.47% | 84.72% |
| 142 | GPT-5.5 | 70.85% | 90.28% | 89.37% |
| 143 | ByteDance Seed 2.0 Lite | 70.73% | 89.50% | 84.27% |
| 144 | GPT-4.1 Nano | 70.27% | 90.07% | 69.90% |
| 145 | Qwen 3.5 397B A17B | 70.17% | 90.04% | 91.09% |
| 146 | Claude Sonnet 4 | 70.11% | 90.04% | 87.64% |
| 147 | Gemini 3.5 Flash (Reasoning, Minimal) | 68.64% | 89.55% | 85.88% |
| 148 | Mistral Medium 3.5 (Reasoning) | 68.10% | 89.37% | 85.32% |
| 149 | DeepSeek V4 Pro | 68.09% | 89.36% | 82.05% |
| 150 | GPT-6 Sol | 67.97% | 89.32% | 90.05% |
| 151 | Qwen 3 32B | 67.79% | 89.06% | 79.37% |
| 152 | Cydonia 24B V4.1 | 66.97% | 85.22% | 72.68% |
| 153 | Mistral Medium 3.1 | 66.78% | 88.93% | 76.08% |
| 154 | Claude Sonnet 5 | 66.23% | 88.74% | 87.34% |
| 155 | Arcee AI: Trinity Mini | 65.05% | 88.35% | 67.68% |
| 156 | Gemini 2.5 Flash Lite | 64.28% | 88.08% | 79.91% |
| 157 | Gemini 3.1 Flash Lite | 63.57% | 87.86% | 85.09% |
| 158 | Claude Opus 4 | 63.52% | 87.84% | 87.22% |
| 159 | Grok 4.3 | 63.41% | 87.79% | 78.00% |
| 160 | Claude Sonnet 4.5 | 63.35% | 87.78% | 87.54% |
| 161 | DeepSeek V4 Flash | 63.20% | 87.73% | 82.02% |
| 162 | Gemini 3.1 Flash Lite (Reasoning) | 62.89% | 87.63% | 85.91% |
| 163 | Mistral Medium 3.5 | 62.75% | 87.58% | 84.52% |
| 164 | Gemini 3.5 Flash Lite (Reasoning, Minimal) | 62.72% | 87.57% | 84.45% |
| 165 | Gemini 3.5 Flash Lite (Reasoning) | 62.47% | 87.49% | 84.34% |
| 166 | Mistral Small 3.2 24B | 62.38% | 86.84% | 77.36% |
| 167 | Gemini 3.1 Flash Lite (Preview) | 61.89% | 87.29% | 85.41% |
| 168 | Llama 3.1 70B | 61.33% | 80.47% | 77.41% |
| 169 | Hermes 3 405B | 61.06% | 87.02% | 80.80% |
| 170 | Cohere Command R+ (Aug. 2024) | 61.00% | 66.30% | 67.04% |
| 171 | GPT-5.4 Nano | 60.58% | 86.86% | 72.16% |
| 172 | Qwen 3.5 Plus (2026-02-15) | 60.20% | 86.62% | 86.17% |
| 173 | GPT-5.4 | 58.96% | 86.32% | 84.31% |
| 174 | DeepSeek V3.2 | 58.75% | 86.25% | 82.22% |
| 175 | GPT-4.1 Mini | 58.70% | 86.23% | 81.40% |
| 176 | Qwen 2.5 72B | 58.36% | 85.54% | 73.17% |
| 177 | GPT-4o, Aug. 6th (temp=1) | 58.17% | 86.00% | 81.28% |
| 178 | DeepSeek V3 (2024-12-26) | 57.07% | 85.69% | 82.62% |
| 179 | WizardLM 2 8x22b | 57.02% | 79.28% | 71.45% |
| 180 | DeepSeek-V2 Chat | 56.98% | 82.48% | 84.09% |
| 181 | Gemini 3 Flash (Preview) | 56.88% | 85.61% | 85.47% |
| 182 | DeepSeek V3.1 | 56.76% | 85.55% | 82.35% |
| 183 | Gemini 2.5 Flash | 56.72% | 85.27% | 80.61% |
| 184 | Mistral Large 2 | 56.35% | 85.45% | 81.50% |
| 185 | GPT-4o Mini (temp=1) | 55.83% | 85.28% | 77.82% |
| 186 | Mistral Large 3 | 55.54% | 85.18% | 84.29% |
| 187 | Hermes 3 70B | 53.99% | 70.71% | 69.74% |
| 188 | GPT-4o, Aug. 6th (temp=0) | 50.00% | 83.33% | 82.18% |
| 189 | Ministral 3 14B | 49.57% | 83.19% | 70.45% |
| 190 | Writer: Palmyra X5 | 48.62% | 80.49% | 78.11% |
| 191 | DeepSeek V3 (2025-03-24) | 47.98% | 76.88% | 79.93% |
| 192 | GPT-4o Mini (temp=0) | 46.70% | 82.23% | 76.86% |
| 193 | GPT-5.4 Mini | 45.55% | 81.85% | 80.45% |
| 194 | Mistral Small 4 | 45.50% | 81.83% | 75.23% |
| 195 | Qwen 3 235B A22B Instruct 2507 | 43.34% | 76.34% | 78.07% |
| 196 | Gemma 3 27B | 41.59% | 80.53% | 75.70% |
| 197 | Grok 4.20 | 41.45% | 80.48% | 81.21% |
| 198 | Ministral 3 3B | 36.00% | 75.10% | 65.02% |
| 199 | Ministral 3B | 35.66% | 73.79% | 59.25% |
| 200 | Mistral NeMO | 31.81% | 69.32% | 63.80% |
| 201 | Gemma 3 12B | 31.38% | 77.13% | 76.07% |
| 202 | Gemma 3 4B | 27.83% | 75.94% | 66.33% |