Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

SimpleBench

A multiple-choice text benchmark designed to test basic reasoning capabilities where non-specialized humans (high school level) consistently outperform state-of-the-art language models, covering spatio-temporal reasoning, social intelligence, and linguistic adversarial robustness.
Source:

Model Performance

#1
Gemini-2.5-Pro-0605
62.4%
#2
Claude-4.1-Opus-Thinking
60.0%
#3
Claude-4.1-Opus
60.0%
#4
Claude-4.0-Opus-Thinking
58.8%
#5
GPT-5
56.7%
#6
GPT-5-Thinking
56.7%
#7
OpenAI-O3-high
53.1%
#8
Gemini-2.5-Pro-Thinking
51.6%
#9
Gemini-2.5-Pro-0325
51.6%
#10
Claude-3.7-Sonnet-Thinking
46.4%
#11
Claude-4.0-Sonnet-Thinking
45.5%
#12
Claude-3.7-Sonnet
44.9%
#13
OpenAI-O1-1217
40.1%
#14
OpenAI-O4-mini-high
38.7%
#15
GPT-4.5
34.5%
#16
Qwen-3
31.0%
#17
Qwen-3-Thinking
31.0%
#18
DeepSeek-R1
30.9%
#19
Gemini-2.0-Flash-Thinking-0121
30.7%
#20
Llama-4-Maverick-17B
27.7%
#21
Claude-3.5-Sonnet-1022
27.5%
#22
DeepSeek-V3-0324
27.2%
#23
Gemini-1.5-Pro
27.1%
#24
GPT-4.1
27.0%
#25
GPT-4o-0513
25.1%
#26
Claude-3-Opus
23.5%
#27
Llama-3.1-405B
23.0%
#28
OpenAI-O3-mini-high
22.8%
#29
Mistral-Large-2
22.5%
#30
GPT-OSS-120B
22.1%
#31
Llama-3.3-70B
19.9%
#32
DeepSeek-V3
18.9%
#33
Gemini-2.0-Flash
18.9%
#34
OpenAI-O1-mini
18.1%
#35
Cohere-Command-R-Plus
17.4%
#36
GPT-4-mini
10.7%