Skip to content

Benchmark

Vals Science Index

Scores are percentages.

Models measured
27
Scores recorded
27
Sources
2
IQ weight
2.15

Score distribution

44.4%
Mean
41.5%
Median
14%
Std dev
28.4%
10th percentile
63.7%
90th percentile
22%
Min
22%68.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.15 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
1GPT-6.2 Astra· Vals AI
Values from 2 sources
68.4%
2GPT-6 Astra· Vals AI65.1%
3Claude Fable 5.2· Vals AI63.7%
4GPT-6.1 Sol· Vals AI59.6%
5Claude Sonnet 5.5· Vals AI59.1%
6Claude Opus 5.5· Vals AI58.7%
7GPT-6.2 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim
Values from 2 sources
58.5%
8Gemini 4 Argon· Vals AI55.4%
9Muse Wonder Full· Vals AI48.5%
10Claude Opus 5· Vals AI47.9%
11GPT-6 Sol· Vals AI45%
12Kimi K3.1 Pro· Vals AI44.4%
13GPT-5.6 Sol· Vals AI41.5%
14Muse Wonder· Vals AI39.1%
15Claude Haiku 5.5· Vals AI38.5%
16Grok 4.6· Vals AI38.1%
17Gemini 3.8 Flash· Vals AI35.8%
18Grok 4.7· Vals AI34.8%
19Inkling 2 Small· Vals AI32.7%
20Ember-1· Vals AI31.8%
21MiMo V2.6 Flash· Vals AI31.7%
22DeepSeek V4.1 Flash· Vals AI29.7%
23Step 5 Preview· Vals AI29.6%
24GPT-6 Luna· Vals AI28.4%
25Gemini 3.1 Pro Preview· Vals AI27.4%
26GPT-5.6 Luna· Vals AI25.3%
27Beam 1.0 501B A23B· Vals AI22%

Data credits: Data by LLM Stats