Skip to content

Benchmark

Blueprint Bench 2 Raw score

Scores are percentages.

Models measured
40
Scores recorded
40
Sources
1
IQ weight
4.03

Score distribution

66.4%
Mean
68.9%
Median
12.2%
Std dev
53.4%
10th percentile
79%
90th percentile
17%
Min
17%82.2%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 4.03 for chat models, measured over 40 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-6.2 Astra· Epoch AI Benchmarking Hub data82.2%
2Gemini 4 Carbon· Epoch AI Benchmarking Hub data80.9%
3GPT-6.2 Sol· Epoch AI Benchmarking Hub data80.2%
4Claude Fable 5.2· Epoch AI Benchmarking Hub data80%
5Gemini 4 Argon· Epoch AI Benchmarking Hub data79%
6Claude Opus 5.5· Epoch AI Benchmarking Hub data77.5%
7GPT-6 Astra· Epoch AI Benchmarking Hub data76.8%
8GPT-6.1 Sol· Epoch AI Benchmarking Hub data73.8%
9GPT-6.1 Terra· Epoch AI Benchmarking Hub data73.3%
10Claude Fable 5.1· Epoch AI Benchmarking Hub data73.2%
11Claude Sonnet 5.5· Epoch AI Benchmarking Hub data72%
12Claude Fable 5· Epoch AI Benchmarking Hub data71.7%
13Gemini 3.8 Flash· Epoch AI Benchmarking Hub data71.7%
14GPT-6 Sol· Epoch AI Benchmarking Hub data70.9%
15GPT 5.5· Epoch AI Benchmarking Hub data70.6%
16Gemini 3.5 Flash· Epoch AI Benchmarking Hub data69.4%
17GPT-5.6 Sol· Epoch AI Benchmarking Hub data69.4%
18Grok 4.6· Epoch AI Benchmarking Hub data69.2%
19Claude Haiku 5.5· Epoch AI Benchmarking Hub data69.1%
20Grok 4.7· Epoch AI Benchmarking Hub data68.9%
21Muse Wonder 1.0· Epoch AI Benchmarking Hub data68.5%
22Gemini 3.6 Flash· Epoch AI Benchmarking Hub data68.3%
23GPT-6 Luna· Epoch AI Benchmarking Hub data68.3%
24GPT-5.6 Terra· Epoch AI Benchmarking Hub data68.1%
25Claude Opus 5· Epoch AI Benchmarking Hub data67.9%
26Kimi K3· Epoch AI Benchmarking Hub data67.5%
27Grok 4.5· Epoch AI Benchmarking Hub data66.5%
28GPT 5.4· Epoch AI Benchmarking Hub data66.4%
29Gemini 3.1 Pro Preview· Epoch AI Benchmarking Hub data66.1%
30Claude Sonnet 5· Epoch AI Benchmarking Hub data65.4%
31Claude Opus 4.7· Epoch AI Benchmarking Hub data65.2%
32GPT-5.6 Luna· Epoch AI Benchmarking Hub data64.3%
33Claude Opus 4.8· Epoch AI Benchmarking Hub data60.6%
34Claude Sonnet 4.6· Epoch AI Benchmarking Hub data57%
35Kimi K2.6· Epoch AI Benchmarking Hub data55.7%
36Gemini 3 Flash· Epoch AI Benchmarking Hub data53.4%
37Grok 4.3· Epoch AI Benchmarking Hub data47.7%
38Gemini Robotics-ER 1.6 Preview· Epoch AI Benchmarking Hub data47.5%
39Claude Haiku 4.5· Epoch AI Benchmarking Hub data36.7%
40Grok 4.20· Epoch AI Benchmarking Hub data17%

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