Skip to content

Benchmark

Vals GPQA Diamond

Science benchmarks measure how well models answer science questions. Scores are percentages. Higher is better.

Models measured
121
Scores recorded
126
Sources
1
IQ weight
2.35

Score distribution

76%
Mean
81.6%
Median
16.7%
Std dev
50.8%
10th percentile
92.7%
90th percentile
27%
Min
27%95.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.35 for chat models, measured over 121 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51GLM 5.1· Vals AI84.5%
52Grok 4.1 Fast· Vals AI84.3%
53Kimi K2.5thinking· Vals AI84.1%
54OpenAI o3· Vals AI84.1%
55Gemini 3.5 Flash Lite· Vals AI83.8%
56Inkling Small· Vals AI83.6%
57GLM 5thinking· Vals AI83.3%
58GPT-5.4 mini· Vals AI83.1%
59Qwen3.5 Flash· Vals AI82.8%
60MiMo V2.5 Pro· Vals AI82.6%
61MiniMax M2.5· Vals AI82.1%
62Claude Sonnet 4.5· Vals AI81.6%
63gemini-2.5-flash-preview-09-2025· Vals AI81.6%
64MiMo V2.5· Vals AI81.6%
65Gemini 3.1 Flash Lite· Vals AI81.1%
66Gemini 2.5 Pro· Vals AI80.8%
67DeepSeek V3.2· Vals AI80.3%
68GPT-5 mini· Vals AI80.3%
69GLM 4.7· Vals AI80.1%
70Grok 3 Mini Fast Betahigh, reasoning· Vals AI79.3%
71GPT OSS 120B· Vals AI78.5%
72Kimi K2 Thinking· Vals AI78.5%
73MiniMax M2.1· Vals AI78.5%
74Qwen3 Max Preview· Vals AI77.8%
75GPT 5.4 Nano· Vals AI77.5%
76gemini-2.5-flash-preview-09-2025thinking· Vals AI76.5%
77Claude Opus 4.1· Vals AI76.3%
78o3 Mini· Vals AI75.5%
79Claude Sonnet 3.7thinking· Vals AI75.3%
80Claude Sonnet 4thinking· Vals AI75%
81GLM 4.6· Vals AI74.5%
82o4 Mini· Vals AI74.5%
83Grok 3· Vals AI74.2%
84o1· Vals AI73.2%
85Grok 3 Mini Fast Betalow, reasoning· Vals AI73%
86Claude Haiku 4.5thinking· Vals AI72.2%
87GLM 4.5· Vals AI72.2%
88Claude Opus 4· Vals AI71.7%
89Kimi K2 Instruct· Vals AI71.5%
90Gemini 2.5 Flash Lite Preview 09-2025thinking· Vals AI70.2%
91Qwen 3 235B A22B· Vals AI70.2%
92Claude Sonnet 4· Vals AI69.4%
93Llama 4 Maverick· Vals AI69.4%
94GPT OSS 20b· Vals AI68.9%
95Mistral Large 3· Vals AI68.4%
96GPT-4.1 mini· Vals AI67.9%
97Claude Sonnet 3.7· Vals AI67.4%
98GPT 4.1· Vals AI65.4%
99Gemini 2.0 Flash· Vals AI65.2%
100Gemini 2.5 Flash Lite Preview 09-2025· Vals AI64.1%

Showing 51–100 of 126 results.