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
1Claude Fable 5.2· Epoch AI Benchmarking Hub data99.2%
2Claude Fable 5.2xhigh· Epoch AI Benchmarking Hub data98.9%
3Claude Opus 5.5max· Epoch AI Benchmarking Hub data98.9%
4Claude Fable 5max· Epoch AI Benchmarking Hub data98.4%
5Gemini 4 Carbonmedium· Epoch AI Benchmarking Hub data98.4%
6Claude Fable 5.2high· Epoch AI Benchmarking Hub data98.1%
7Claude Fable 5.2low· Epoch AI Benchmarking Hub data98.1%
8Claude Fable 5.2medium· Epoch AI Benchmarking Hub data98.1%
9Gemini 4 Carbon· Epoch AI Benchmarking Hub data98.1%
10Claude Opus 5high· Epoch AI Benchmarking Hub data97.9%
11Claude Opus 5xhigh· Epoch AI Benchmarking Hub data97.9%
12Claude Fable 5.1· Epoch AI Benchmarking Hub data97.6%
13Claude Fable 5.1xhigh· Epoch AI Benchmarking Hub data97.6%
14Claude Opus 5max· Epoch AI Benchmarking Hub data97.6%
15Gemini 4 Argonlow· Epoch AI Benchmarking Hub data97.6%
16Gemini 4 Carbonlow· Epoch AI Benchmarking Hub data97.6%
17Claude Fable 5xhigh· Epoch AI Benchmarking Hub data97.3%
18Claude Opus 5.5xhigh· Epoch AI Benchmarking Hub data97.3%
19Claude Sonnet 5.5max· Epoch AI Benchmarking Hub data97.3%
20Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data97.3%
21Gemini 4 Argonmedium· Epoch AI Benchmarking Hub data97.3%
22GPT-6 Astraxhigh· Epoch AI Benchmarking Hub data97.3%
23GPT-6 Solmax· Epoch AI Benchmarking Hub data97.3%
24Grok 4.6xhigh· Epoch AI Benchmarking Hub data97.3%
25Claude Fable 5.1medium· Epoch AI Benchmarking Hub data97.1%
26Gemini 3.1 Pro Previewmedium· Epoch AI Benchmarking Hub data97.1%
27GPT-6 Astra· Epoch AI Benchmarking Hub data97.1%
28GPT-6 Astramedium· Epoch AI Benchmarking Hub data97.1%
29Gemini 3.7 Flashhigh· Epoch AI Benchmarking Hub data96.8%
30Claude Fable 5high· Epoch AI Benchmarking Hub data96.5%
31Claude Fable 5.1high· Epoch AI Benchmarking Hub data96.5%
32Claude Opus 5.5high· Epoch AI Benchmarking Hub data96.5%
33Claude Opus 5.5medium· Epoch AI Benchmarking Hub data96.5%
34GPT-6 Astrahigh· Epoch AI Benchmarking Hub data96.5%
35GPT-6.1 Solmax· Epoch AI Benchmarking Hub data96.5%
36Grok 4.5high· Epoch AI Benchmarking Hub data96.5%
37Muse Spark 1.3· Epoch AI Benchmarking Hub data96.5%
38Claude Sonnet 5.5xhigh· Epoch AI Benchmarking Hub data96%
39GPT 5.5xhigh· Epoch AI Benchmarking Hub data96%
40GPT-5.5 Proxhigh· Epoch AI Benchmarking Hub data96%
41GPT-5.6 Solpro / max· Epoch AI Benchmarking Hub data96%
42Grok 4.7xhigh· Epoch AI Benchmarking Hub data96%
43Claude Fable 5medium· Epoch AI Benchmarking Hub data95.7%
44GPT-5.6 Solxhigh· Epoch AI Benchmarking Hub data95.7%
45Gemini 3.8 Flash· Epoch AI Benchmarking Hub data95.7%
46Claude Opus 5.5low· Epoch AI Benchmarking Hub data95.5%
47Gemini 3.1 Pro Previewhigh· Epoch AI Benchmarking Hub data95.5%
48Gemini 3.6 Flashhigh· Epoch AI Benchmarking Hub data95.5%
49GPT-5.6 Solhigh· Epoch AI Benchmarking Hub data95.5%
50GPT-5.6 Solmax· Epoch AI Benchmarking Hub data95.5%

Showing 1–50 of 239 results.

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