Skip to content

Benchmark

Weirdml V3 Max performance

Scores are percentages.

Models measured
27
Scores recorded
27
Sources
1
IQ weight
2.27

Score distribution

26.1%
Mean
21.4%
Median
16.1%
Std dev
9.7%
10th percentile
49.8%
90th percentile
4.4%
Min
4.4%65.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.27 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
1Claude Fable 5.2xhigh· Epoch AI Benchmarking Hub data65.3%
2GPT-6 Astraxhigh· Epoch AI Benchmarking Hub data54%
3GPT-6.1 Solxhigh· Epoch AI Benchmarking Hub data50.6%
4Claude Opus 5.5xhigh· Epoch AI Benchmarking Hub data49.8%
5Claude Fable 5.1xhigh· Epoch AI Benchmarking Hub data43.5%
6Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data41.4%
7GPT-6.1 Terraxhigh· Epoch AI Benchmarking Hub data40.4%
8Claude Sonnet 5.5xhigh· Epoch AI Benchmarking Hub data37%
9Claude Opus 5xhigh· Epoch AI Benchmarking Hub data32.5%
10Claude Fable 5xhigh· Epoch AI Benchmarking Hub data27.9%
11GPT-6 Solxhigh· Epoch AI Benchmarking Hub data27.6%
12Muse Wonder 1.0xhigh· Epoch AI Benchmarking Hub data24.9%
13Claude Haiku 5.5xhigh· Epoch AI Benchmarking Hub data24.1%
14GPT-5.6 Solxhigh· Epoch AI Benchmarking Hub data21.4%
15DeepSeek V4.1 Flash 1017high· Epoch AI Benchmarking Hub data19.7%
16Kimi K3.1 Proxhigh· Epoch AI Benchmarking Hub data18.9%
17Grok 4.7xhigh· Epoch AI Benchmarking Hub data18.7%
18Kimi K3high· Epoch AI Benchmarking Hub data15.4%
19Gemini 3.8 Flashhigh· Epoch AI Benchmarking Hub data14.6%
20Muse Spark 1.3xhigh· Epoch AI Benchmarking Hub data12.1%
21GPT-6 Lunaxhigh· Epoch AI Benchmarking Hub data12%
22Mistral Large 4high· Epoch AI Benchmarking Hub data11.8%
23DeepSeek V4.1 Flashhigh· Epoch AI Benchmarking Hub data11.6%
24GPT 5.2xhigh· Epoch AI Benchmarking Hub data9.7%
25GPT-5.6 Lunaxhigh· Epoch AI Benchmarking Hub data9.3%
26Claude Opus 4.5high· Epoch AI Benchmarking Hub data7.1%
27Inkling FP4xhigh· Epoch AI Benchmarking Hub data4.4%

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)