Skip to content

Benchmark

Vals Index

Scores are percentages.

Models measured
40
Scores recorded
40
Sources
2
IQ weight
2.93

Score distribution

57.5%
Mean
59.4%
Median
11%
Std dev
43.7%
10th percentile
70.1%
90th percentile
27.6%
Min
27.6%71.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.93 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
1Claude Fable 5.2· Vals AI71.3%
2Claude Sonnet 5.5· Vals AI70.9%
3Gemini 4 Carbon· Vals AI
Values from 2 sources
70.5%
4Claude Opus 5.5· Vals AI70.1%
5Gemini 4 Argon· Vals AI
Values from 2 sources
69.4%
6Gemini 4 Hydrogen· Vals AI68.5%
7GPT-6.2 Astra· Vals AI67.7%
8Muse Wonder Full· Vals AI66.8%
9GPT-6.2 Sol· Vals AI66.1%
10GPT-6 Astra· Vals AI65%
11Muse Wonder· Vals AI64.9%
12GPT-6.1 Sol· Vals AI63.9%
13Qwen4 Max· Vals AI63.2%
14Kimi K3.1 Pro· Vals AI62.5%
15DeepSeek V4.1 Flash 1017· Vals AI61.3%
16GPT-6.1 Terra· Vals AI60.5%
17Muse Ember· Vals AI60.5%
18GLM 5.3· Vals AI59.8%
19Hy 4· Vals AI59.4%
20Claude Haiku 5.5· Vals AI58.7%
21Grok 4.7· Vals AI58.6%
22MiMo V2.6 Pro· Vals AI57.8%
23Inkling 2 Small· Vals AI56.1%
24Gemini 3.8 Flash· Vals AI56%
25MiMo V2.6 Flash· Vals AI54.8%
26Step 5 Preview· Vals AI54.8%
27MiniMax M3.1 Flash· Vals AI53.8%
28Hy4 preview· Vals AI53.2%
29DeepSeek V4 Pro 0813· Vals AI53%
30Kimi K3· Vals AI53%
31GPT-6 Luna· Vals AI52.3%
32GLM 5.3 Flash· Vals AI50.9%
33Hy4 Flash· Vals AI50.8%
34Muse Spark Open· Vals AI48.6%
35Inkling 2 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim46.5%
36Mercury 3· Vals AI43.7%
37Beam 1.0 501B A23B· Vals AI43.4%
38NVIDIA Nemotron 3.5 Super VL 120B· Vals AI29.8%
39Inkling FP4· Vals AI29%
40Inkling Small· Vals AI27.6%

Data credits: Data by LLM Stats