Skip to content

Benchmark

Dtbench Accuracy

Scores are percentages. Higher is better.

Models measured
170
Scores recorded
239
Sources
1
IQ weight
1.22

Score distribution

79.8%
Mean
84.5%
Median
16.6%
Std dev
52.5%
10th percentile
97.1%
90th percentile
41.7%
Min
41.7%99.2%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.22 for chat models, measured over 170 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101Gemini 3 Flashhigh· Epoch AI Benchmarking Hub data89.1%
102GPT-5.6 Lunaxhigh· Epoch AI Benchmarking Hub data89.1%
103GPT-5.6 Lunamax· Epoch AI Benchmarking Hub data88.8%
104Mistral Large 4· Epoch AI Benchmarking Hub data88.2%
105Qwen 3.8 27B· Epoch AI Benchmarking Hub data88%
106DeepSeek V3.2 Expthinking· Epoch AI Benchmarking Hub data87.7%
107Grok 4.1 Fastreasoning· Epoch AI Benchmarking Hub data87.7%
108GLM 5.3· Epoch AI Benchmarking Hub data87.7%
109Inkling FP4xhigh· Epoch AI Benchmarking Hub data87.5%
110Qwen3.5 397B A17B· Epoch AI Benchmarking Hub data87.5%
111Claude Haiku 5.5xhigh· Epoch AI Benchmarking Hub data87.2%
112Qwen3.6 Max Preview· Epoch AI Benchmarking Hub data87.2%
113O3 Prohigh· Epoch AI Benchmarking Hub data86.9%
114Claude Haiku 5.5high· Epoch AI Benchmarking Hub data86.7%
115GPT-5.6 Lunahigh· Epoch AI Benchmarking Hub data86.7%
116DeepSeek V4 Flashmax· Epoch AI Benchmarking Hub data86.4%
117DeepSeek V3.2· Epoch AI Benchmarking Hub data85.6%
118OpenAI o3high· Epoch AI Benchmarking Hub data84.8%
119Claude Sonnet 5high· Epoch AI Benchmarking Hub data84.5%
120MiMo V2.5 Pro· Epoch AI Benchmarking Hub data84.5%
121Claude Haiku 5.5medium· Epoch AI Benchmarking Hub data84.3%
122Qwen3.5 122B-A10Bnone· Epoch AI Benchmarking Hub data84.3%
123Qwen3.7 Plus· Epoch AI Benchmarking Hub data84%
124GPT-5.6 Solnone· Epoch AI Benchmarking Hub data83.7%
125Gemini 3.5 Flash Litehigh· Epoch AI Benchmarking Hub data83.5%
126Claude Sonnet 4.5· Epoch AI Benchmarking Hub data83.2%
127GPT-5.6 Lunamedium· Epoch AI Benchmarking Hub data83.2%
128Qwen3.5 Flash· Epoch AI Benchmarking Hub data82.9%
129DeepSeek V3.1thinking· Epoch AI Benchmarking Hub data82.7%
130Gemma 4 31B IT· Epoch AI Benchmarking Hub data82.7%
131Grok 4 Fast· Epoch AI Benchmarking Hub data82.7%
132Gemini 2.5 Pro· Epoch AI Benchmarking Hub data82.4%
133Qwen3.5 27B· Epoch AI Benchmarking Hub data82.4%
134Qwen3 Max· Epoch AI Benchmarking Hub data82.1%
135Qwen3.6 Plus· Epoch AI Benchmarking Hub data81.9%
136Claude Opus 4· Epoch AI Benchmarking Hub data81.6%
137DeepSeek V3.1 Terminus· Epoch AI Benchmarking Hub data81.3%
138Claude Haiku 5.5low· Epoch AI Benchmarking Hub data81.1%
139Qwen Plus· Epoch AI Benchmarking Hub data81.1%
140GPT-5 minihigh· Epoch AI Benchmarking Hub data80.5%
141Qwen3.5 Plus· Epoch AI Benchmarking Hub data80.5%
142GPT 5.4 Nanoxhigh· Epoch AI Benchmarking Hub data80.3%
143Qwen 3 235B A22B Thinking 2507· Epoch AI Benchmarking Hub data80.3%
144Claude Opus 4.1· Epoch AI Benchmarking Hub data80%
145GPT-5.4 minixhigh· Epoch AI Benchmarking Hub data80%
146GPT-5.6 Lunalow· Epoch AI Benchmarking Hub data80%
147Qwen3.5 35B A3B· Epoch AI Benchmarking Hub data80%
148Claude Sonnet 5medium· Epoch AI Benchmarking Hub data79.2%
149MiniMax M3· Epoch AI Benchmarking Hub data78.9%
150GPT-5.6 Terranone· Epoch AI Benchmarking Hub data78.7%

Showing 101–150 of 239 results.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)