Skip to content

Benchmark

Dtbench CRI-rescaled score

Scores are percentages.

Models measured
170
Scores recorded
239
Sources
1
IQ weight
1.22

Score distribution

66.3%
Mean
74.2%
Median
27.6%
Std dev
20.9%
10th percentile
95.1%
90th percentile
2.8%
Min
2.8%98.7%
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 data81.8%
102GPT-5.6 Lunaxhigh· Epoch AI Benchmarking Hub data81.8%
103GPT-5.6 Lunamax· Epoch AI Benchmarking Hub data81.3%
104Mistral Large 4· Epoch AI Benchmarking Hub data80.4%
105Qwen 3.8 27B· Epoch AI Benchmarking Hub data80%
106Grok 4.1 Fastreasoning· Epoch AI Benchmarking Hub data79.6%
107DeepSeek V3.2 Expthinking· Epoch AI Benchmarking Hub data79.6%
108GLM 5.3· Epoch AI Benchmarking Hub data79.5%
109Inkling FP4xhigh· Epoch AI Benchmarking Hub data79.1%
110Qwen3.5 397B A17B· Epoch AI Benchmarking Hub data79.1%
111Claude Haiku 5.5xhigh· Epoch AI Benchmarking Hub data78.7%
112Qwen3.6 Max Preview· Epoch AI Benchmarking Hub data78.7%
113O3 Prohigh· Epoch AI Benchmarking Hub data78.2%
114Claude Haiku 5.5high· Epoch AI Benchmarking Hub data77.8%
115GPT-5.6 Lunahigh· Epoch AI Benchmarking Hub data77.8%
116DeepSeek V4 Flashmax· Epoch AI Benchmarking Hub data77.3%
117DeepSeek V3.2· Epoch AI Benchmarking Hub data76%
118OpenAI o3high· Epoch AI Benchmarking Hub data74.7%
119Claude Sonnet 5high· Epoch AI Benchmarking Hub data74.2%
120MiMo V2.5 Pro· Epoch AI Benchmarking Hub data74.2%
121Claude Haiku 5.5medium· Epoch AI Benchmarking Hub data73.8%
122Qwen3.5 122B-A10Bnone· Epoch AI Benchmarking Hub data73.8%
123Qwen3.7 Plus· Epoch AI Benchmarking Hub data73.3%
124GPT-5.6 Solnone· Epoch AI Benchmarking Hub data72.9%
125Gemini 3.5 Flash Litehigh· Epoch AI Benchmarking Hub data72.5%
126Claude Sonnet 4.5· Epoch AI Benchmarking Hub data72%
127GPT-5.6 Lunamedium· Epoch AI Benchmarking Hub data72%
128Qwen3.5 Flash· Epoch AI Benchmarking Hub data71.6%
129DeepSeek V3.1thinking· Epoch AI Benchmarking Hub data71.1%
130Gemma 4 31B IT· Epoch AI Benchmarking Hub data71.1%
131Grok 4 Fast· Epoch AI Benchmarking Hub data71.1%
132Gemini 2.5 Pro· Epoch AI Benchmarking Hub data70.7%
133Qwen3.5 27B· Epoch AI Benchmarking Hub data70.7%
134Qwen3 Max· Epoch AI Benchmarking Hub data70.2%
135Qwen3.6 Plus· Epoch AI Benchmarking Hub data69.8%
136Claude Opus 4· Epoch AI Benchmarking Hub data69.3%
137DeepSeek V3.1 Terminus· Epoch AI Benchmarking Hub data68.9%
138Claude Haiku 5.5low· Epoch AI Benchmarking Hub data68.5%
139Qwen Plus· Epoch AI Benchmarking Hub data68.5%
140GPT-5 minihigh· Epoch AI Benchmarking Hub data67.6%
141Qwen3.5 Plus· Epoch AI Benchmarking Hub data67.6%
142GPT 5.4 Nanoxhigh· Epoch AI Benchmarking Hub data67.1%
143Qwen 3 235B A22B Thinking 2507· Epoch AI Benchmarking Hub data67.1%
144Claude Opus 4.1· Epoch AI Benchmarking Hub data66.7%
145GPT-5.4 minixhigh· Epoch AI Benchmarking Hub data66.7%
146GPT-5.6 Lunalow· Epoch AI Benchmarking Hub data66.7%
147Qwen3.5 35B A3B· Epoch AI Benchmarking Hub data66.7%
148Claude Sonnet 5medium· Epoch AI Benchmarking Hub data65.3%
149MiniMax M3· Epoch AI Benchmarking Hub data64.9%
150GPT-5.6 Terranone· Epoch AI Benchmarking Hub data64.5%

Showing 101–150 of 239 results.

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