Skip to content

Benchmark

NYT Connections – Newest 100 puzzles, extended version (Lech Mazur)

Scores are percentages.

Models measured
106
Scores recorded
138
Sources
1
IQ weight
2.05

Score distribution

61.5%
Mean
70.9%
Median
30%
Std dev
14.6%
10th percentile
94.5%
90th percentile
5.5%
Min
5.5%99.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.05 for chat models, measured over 106 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101Claude Opus 4.7high reasoning· Lech Mazur suite data39.3%
102Claude Haiku 4.5· Lech Mazur suite data37.5%
103Claude Opus 4.6no reasoning· Lech Mazur suite data37.2%
104Qwen3.6 35B A3B· Lech Mazur suite data35.5%
105Qwen3.7 Flash· Lech Mazur suite data35.4%
106Step 3.7 Flashhigh reasoning· Lech Mazur suite data34.2%
107Qwen3 Max· Lech Mazur suite data31.6%
108MiMo V2.5 Pro· Lech Mazur suite data31.4%
109DeepSeek V3.2· Lech Mazur suite data29.7%
110MiniMax M3.1 Flashhigh reasoning· Lech Mazur suite data29.2%
111MiniMax M2.7· Lech Mazur suite data28.6%
112Mistral Large 4high· Lech Mazur suite data27.2%
113Step 3.5 Flash· Lech Mazur suite data22.8%
114MiMo V2 Pro· Lech Mazur suite data22%
115ERNIE 5.1· Lech Mazur suite data20%
116GPT 5.4no reasoning· Lech Mazur suite data19.8%
117GPT 5.5no reasoning· Lech Mazur suite data19.1%
118Hy3 (Deep Infra)· Lech Mazur suite data18.9%
119LongCat-Flash-Thinking· Lech Mazur suite data17.3%
120Trinity Large Thinking· Lech Mazur suite data16.9%
121MiniMax M2.5· Lech Mazur suite data16.2%
122GPT 5.2no reasoning· Lech Mazur suite data15.8%
123Gemma 4 31B IT· Lech Mazur suite data15.7%
124Mistral Medium 3.5high· Lech Mazur suite data14.6%
125MiniMax M2.1· Lech Mazur suite data12.8%
126Nemotron 3 Super· Lech Mazur suite data12.4%
127Claude Opus 4.7no reasoning· Lech Mazur suite data11.9%
128MiniMax M2· Lech Mazur suite data11.9%
129ERNIE 5.0· Lech Mazur suite data10.7%
130Grok 4.20Non-Reasoning· Lech Mazur suite data9.4%
131Gemini 3.1 Flash Lite· Lech Mazur suite data8.9%
132Mistral Large 3· Lech Mazur suite data8.9%
133Grok 4.1 Fastnon-reasoning· Lech Mazur suite data8.8%
134Hy3 (Deep Infra)non-thinking· Lech Mazur suite data8.2%
135Llama 4 Maverick· Lech Mazur suite data8.1%
136Mistral Medium 3.1· Lech Mazur suite data7.1%
137DeepSeek V3.2no reasoning· Lech Mazur suite data6.8%
138ling-2.6-1t· Lech Mazur suite data5.5%

Showing 101–138 of 138 results.