Skip to content

Benchmark

CritPT

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

Models measured
156
Scores recorded
268
Sources
3
IQ weight
2.67

Score distribution

0.98%
Mean
0.106%
Median
4%
Std dev
0%
10th percentile
0.314%
90th percentile
0%
Min
0%31.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.67 for chat models, measured over 156 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 4 Carbonhigh· Epoch AI Benchmarking Hub data36%
2Claude Fable 5.2xhigh· Epoch AI Benchmarking Hub data35.7%
3Claude Fable 5.2max· Epoch AI Benchmarking Hub data35.4%
4Claude Fable 5.2high· Epoch AI Benchmarking Hub data32.9%
5GPT-5.6 Solmax· Epoch AI Benchmarking Hub data32.3%
6Claude Opus 5.5max· Epoch AI Benchmarking Hub data31.7%
7Claude Opus 5.5xhigh· Epoch AI Benchmarking Hub data31.7%
8GPT-6 Astramax· Epoch AI Benchmarking Hub data31.7%
9GPT-6.1 Solmax· Epoch AI Benchmarking Hub data31.7%
10GPT-6.1 Solxhigh· Epoch AI Benchmarking Hub data31.7%
11Claude Sonnet 5.5max· Epoch AI Benchmarking Hub data31.4%
12GPT-6 Astraxhigh· Epoch AI Benchmarking Hub data31.4%
13Inkling 2 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim31.4%
14Claude Fable 5.1xhigh· Epoch AI Benchmarking Hub data31.1%
15Claude Sonnet 5.5xhigh· Epoch AI Benchmarking Hub data31.1%
16Claude Fable 5.2medium· Epoch AI Benchmarking Hub data30.9%
17Claude Opus 5.5high· Epoch AI Benchmarking Hub data30.9%
18GPT-6 Solmax· Epoch AI Benchmarking Hub data30.9%
19GPT-5.5 Proxhigh· Epoch AI Benchmarking Hub data30.6%
20Claude Fable 5.1high· Epoch AI Benchmarking Hub data30.3%
21GPT-5.4 Proxhigh· Epoch AI Benchmarking Hub data30%
22GPT-5.6 Terramax· Epoch AI Benchmarking Hub data30%
23GPT-6.1 Solhigh· Epoch AI Benchmarking Hub data30%
24Claude Fable 5.1max· Epoch AI Benchmarking Hub data29.7%
25Claude Fable 5.2low· Epoch AI Benchmarking Hub data29.4%
26Claude Fable 5.1medium· Epoch AI Benchmarking Hub data29.1%
27Claude Opus 5max· Epoch AI Benchmarking Hub data29.1%
28GPT-6 Astramedium· Epoch AI Benchmarking Hub data29.1%
29GPT-6 Astrahigh· Epoch AI Benchmarking Hub data28.9%
30GPT-5.6 Solxhigh· Epoch AI Benchmarking Hub data28.6%
31Claude Fable 5max· Epoch AI Benchmarking Hub data28.6%
32Claude Opus 5high· Epoch AI Benchmarking Hub data28.3%
33Muse Wonder 1.0max· Epoch AI Benchmarking Hub data28.3%
34GPT-6 Solxhigh· Epoch AI Benchmarking Hub data28%
35GPT-6.1 Terramax· Epoch AI Benchmarking Hub data28%
36Claude Fable 5.1low· Epoch AI Benchmarking Hub data27.7%
37Claude Opus 5xhigh· Epoch AI Benchmarking Hub data27.7%
38Claude Opus 5.5medium· Epoch AI Benchmarking Hub data27.7%
39GPT-6.1 Solmedium· Epoch AI Benchmarking Hub data27.7%
40Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data27.1%
41GPT 5.5xhigh· Epoch AI Benchmarking Hub data27.1%
42GPT-5.6 Terraxhigh· Epoch AI Benchmarking Hub data27.1%
43Claude Opus 4.8Fable/Mythos 5 System Card capability summary · adaptive thinking · max effort · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.27.1%
44Claude Opus 5medium· Epoch AI Benchmarking Hub data26.9%
45GPT-6.1 Terraxhigh· Epoch AI Benchmarking Hub data26.6%
46MiMo V2.6 Pro· Epoch AI Benchmarking Hub data26.6%
47GPT-6 Astralow· Epoch AI Benchmarking Hub data26.3%
48Muse Spark 1.3xhigh· Epoch AI Benchmarking Hub data26%
49GPT-5.6 Solhigh· Epoch AI Benchmarking Hub data25.7%
50GPT-6.1 Terrahigh· Epoch AI Benchmarking Hub data25.7%

Showing 1–50 of 268 results.

Data credits: Data by LLM Stats · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.