Skip to content

Benchmark

Vals TaxEval v2

Scores are percentages.

Models measured
128
Scores recorded
135
Sources
1
IQ weight
2.45

Score distribution

69.4%
Mean
72.1%
Median
9.3%
Std dev
60.3%
10th percentile
75.6%
90th percentile
1.6%
Min
1.6%80.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.45 for chat models, measured over 128 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Grok 4.1 Fast· Vals AI73.1%
52Nemotron 3 Ultra· Vals AI73.1%
53DeepSeek V4 Pro 0813· Vals AI73.1%
54Mistral Large 3· Vals AI73.1%
55Grok 3 Mini Fast Betalow, reasoning· Vals AI73%
56Gemini 2.5 Pro· Vals AI72.9%
57Gemini 3.1 Pro Preview· Vals AI72.9%
58gemini-2.5-flash-preview-09-2025· Vals AI72.7%
59MiniMax M3· Vals AI72.7%
60Gemini 3.5 Flash Lite· Vals AI72.6%
61Gemini 3 Pro Preview· Vals AI72.6%
62Claude Sonnet 3.7· Vals AI72.4%
63gemini-2.5-flash-preview-09-2025thinking· Vals AI72.4%
64GLM 4.5· Vals AI72.4%
65GLM 5.3· Vals AI72.4%
66DeepSeek R1· Vals AI72.3%
67Qwen3.5 Flash· Vals AI72.2%
68DeepSeek V4 Pro· Vals AI72.1%
69Claude Sonnet 4thinking· Vals AI72%
70Claude Opus 4· Vals AI71.9%
71GPT-4.1 mini· Vals AI71.9%
72MiMo V2.5· Vals AI71.8%
73Gemini 3.1 Flash Lite· Vals AI71.8%
74Kimi K2 Thinking· Vals AI71.7%
75Grok 4.5· Vals AI71.7%
76GPT OSS 120B· Vals AI71.6%
77Qwen3.6 27B· Vals AI71.3%
78GPT-5.4 mini· Vals AI71.2%
79GLM 5.1· Vals AI71.2%
80gemini-2.5-flash-preview-04-17· Vals AI71.2%
81Grok 3 Mini Fast Betahigh, reasoning· Vals AI71.1%
82DeepSeek V3· Vals AI71.1%
83Grok 4.6· Vals AI71.1%
84Qwen 3.8 27B· Vals AI70.8%
85Grok 4.3· Vals AI70.8%
86DeepSeek V4 Flash 0731· Vals AI70.7%
87Qwen 3 235B A22B· Vals AI70.6%
88gemini-2.5-flash-preview-04-17thinking· Vals AI70.5%
89Mistral Medium 3· Vals AI70.3%
90Kimi K2 Instruct· Vals AI70.2%
91Claude 3.5 Sonnet· Vals AI70.2%
92GLM 5thinking· Vals AI70%
93Gemini 2.0 Flash Experimentalthinking· Vals AI69.8%
94Claude Sonnet 4· Vals AI69.6%
95o3 Mini· Vals AI69.4%
96GLM 4.7· Vals AI68.8%
97DeepSeek V3.2· Vals AI68.2%
98gemini-2.0-pro-exp-02-05· Vals AI68.2%
99MiniMax M2.5· Vals AI68.2%
100Mistral Medium 3.5· Vals AI68%

Showing 51–100 of 135 results.