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
51GPT-5.6 Solmedium· Epoch AI Benchmarking Hub data95.2%
52Muse Wonder 1.0· Epoch AI Benchmarking Hub data95.2%
53Claude Haiku 5.5max· Epoch AI Benchmarking Hub data95%
54Claude Opus 4.8max· Epoch AI Benchmarking Hub data94.9%
55Claude Opus 5medium· Epoch AI Benchmarking Hub data94.9%
56Gemini 3.1 Pro Previewlow· Epoch AI Benchmarking Hub data94.9%
57Gemini 3.8 Flashlow· Epoch AI Benchmarking Hub data94.9%
58Gemini 3.8 Flashmedium· Epoch AI Benchmarking Hub data94.9%
59GPT-6 Astralow· Epoch AI Benchmarking Hub data94.9%
60Gemini 3.5 Flashhigh· Epoch AI Benchmarking Hub data94.7%
61Gemini 3.7 Flashmedium· Epoch AI Benchmarking Hub data94.7%
62Muse Spark 1.2xhigh· Epoch AI Benchmarking Hub data94.7%
63Claude Opus 4.7max· Epoch AI Benchmarking Hub data94.7%
64GPT 5.4xhigh· Epoch AI Benchmarking Hub data94.4%
65Muse Spark 1.1high· Epoch AI Benchmarking Hub data94.4%
66Claude Fable 5.1low· Epoch AI Benchmarking Hub data94.1%
67Claude Sonnet 5.5high· Epoch AI Benchmarking Hub data94.1%
68Claude Sonnet 5.5medium· Epoch AI Benchmarking Hub data93.9%
69DeepSeek V4 Pro 0813· Epoch AI Benchmarking Hub data93.9%
70GPT-5.6 Sollow· Epoch AI Benchmarking Hub data93.9%
71GLM 5.2max· Epoch AI Benchmarking Hub data93.6%
72Claude Sonnet 5.5low· Epoch AI Benchmarking Hub data93.3%
73Gemini 3.7 Flashlow· Epoch AI Benchmarking Hub data93.3%
74GPT-5.6 Terramax· Epoch AI Benchmarking Hub data93.3%
75Kimi K3.1 Promax· Epoch AI Benchmarking Hub data93.3%
76Claude Opus 5low· Epoch AI Benchmarking Hub data93.1%
77GPT-5.6 Terraxhigh· Epoch AI Benchmarking Hub data92.5%
78Claude Sonnet 5max· Epoch AI Benchmarking Hub data92.5%
79Qwen3.7 Max· Epoch AI Benchmarking Hub data92.3%
80Qwen3.8 Maxxhigh· Epoch AI Benchmarking Hub data92%
81Claude Fable 5low· Epoch AI Benchmarking Hub data91.5%
82Claude Opus 4.6max· Epoch AI Benchmarking Hub data91.2%
83GPT-5.6 Terrahigh· Epoch AI Benchmarking Hub data91.2%
84Kimi K3max· Epoch AI Benchmarking Hub data91.2%
85DeepSeek V4 Flash 0731max· Epoch AI Benchmarking Hub data90.9%
86GPT 5.2xhigh· Epoch AI Benchmarking Hub data90.9%
87Kimi K2.6· Epoch AI Benchmarking Hub data90.9%
88DeepSeek V4 Promax· Epoch AI Benchmarking Hub data90.7%
89GPT 5high· Epoch AI Benchmarking Hub data90.7%
90Grok 4.3high· Epoch AI Benchmarking Hub data90.7%
91GPT 5.1high· Epoch AI Benchmarking Hub data90.1%
92GPT-5.6 Terralow· Epoch AI Benchmarking Hub data90.1%
93GPT-5.6 Terramedium· Epoch AI Benchmarking Hub data90.1%
94GPT-6 Lunamax· Epoch AI Benchmarking Hub data90.1%
95Grok 4.20· Epoch AI Benchmarking Hub data90.1%
96Nemotron 3 Ultra· Epoch AI Benchmarking Hub data90.1%
97Claude Opus 4.5high· Epoch AI Benchmarking Hub data89.9%
98Claude Sonnet 4.6max· Epoch AI Benchmarking Hub data89.9%
99DeepSeek V4.1 Flash· Epoch AI Benchmarking Hub data89.9%
100Claude Sonnet 5xhigh· Epoch AI Benchmarking Hub data89.5%

Showing 51–100 of 239 results.

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