Skip to content

Benchmark

VPCT

Scores are percentages.

Models measured
27
Scores recorded
37
Sources
1
IQ weight
2.27

Score distribution

45.7%
Mean
40%
Median
14.7%
Std dev
33%
10th percentile
66%
90th percentile
30%
Min
30%91%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.27 for chat models, measured over 27 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 3 Pro Preview· Epoch AI Benchmarking Hub data91%
2GPT 5.2xhigh· Epoch AI Benchmarking Hub data84%
3Gemini 3 Flash· Epoch AI Benchmarking Hub data72.6%
4GPT 5.2high· Epoch AI Benchmarking Hub data67%
5GPT 5high· Epoch AI Benchmarking Hub data66%
6GPT 5medium· Epoch AI Benchmarking Hub data63.2%
7GPT 5.1high· Epoch AI Benchmarking Hub data58.7%
8o4 Minimedium· Epoch AI Benchmarking Hub data57.5%
9GPT 5.1medium· Epoch AI Benchmarking Hub data53.3%
10OpenAI o3medium· Epoch AI Benchmarking Hub data52%
11Gemini 2.5 Pro Preview 03-25· Epoch AI Benchmarking Hub data48%
12Gemini 2.5 Pro· Epoch AI Benchmarking Hub data46.4%
13gemini-2.5-flash-preview-09-2025· Epoch AI Benchmarking Hub data46.2%
14GPT-4.5 Preview· Epoch AI Benchmarking Hub data45%
15gemini-robotics-er-1.5-preview· Epoch AI Benchmarking Hub data40.8%
16Gemini 2.5 Pro Preview 05-06· Epoch AI Benchmarking Hub data40.5%
17GPT-5 minimedium· Epoch AI Benchmarking Hub data40.2%
18Claude Opus 4.532k· Epoch AI Benchmarking Hub data40%
19GPT 4o· Epoch AI Benchmarking Hub data40%
20Claude Sonnet 4.532k· Epoch AI Benchmarking Hub data39.8%
21Claude Sonnet 3.7· Epoch AI Benchmarking Hub data39%
22GPT-5 minihigh· Epoch AI Benchmarking Hub data39%
23Claude Opus 416k· Epoch AI Benchmarking Hub data38%
24Claude Sonnet 4.5· Epoch AI Benchmarking Hub data38%
25gemini-2.5-flash-preview-04-17· Epoch AI Benchmarking Hub data38%
26GPT-5 nanohigh· Epoch AI Benchmarking Hub data37.2%
27o1medium· Epoch AI Benchmarking Hub data37%
28GPT-5 nanomedium· Epoch AI Benchmarking Hub data35.4%
29Claude Opus 4.1· Epoch AI Benchmarking Hub data35%
30Claude Sonnet 3.764k· Epoch AI Benchmarking Hub data35%
31Claude Sonnet 432k· Epoch AI Benchmarking Hub data34%
32GPT-4o mini· Epoch AI Benchmarking Hub data34%
33Claude 3.5 Sonnet· Epoch AI Benchmarking Hub data33%
34Claude Opus 4· Epoch AI Benchmarking Hub data33%
35Claude Opus 4.116k· Epoch AI Benchmarking Hub data33%
36Claude Sonnet 4· Epoch AI Benchmarking Hub data30%
37Gemini 2.5 Flash Lite Preview 09-2025· Epoch AI Benchmarking Hub data30%

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