Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

LiveBench (Data Analysis)

Measures performance on data analysis tasks using recent datasets from Kaggle and Socrata, including table reformatting, join prediction, and column type annotation.
Source:

Model Performance

#1
GPT-5-mini
71.9%
#2
GPT-5
71.6%
#3
Gemini-2.5-Pro-0325
71.6%
#4
DeepSeek-R1
71.5%
#5
Claude-4.1-Opus-Thinking
71.1%
#6
Claude-4.0-Opus-Thinking
70.7%
#7
Gemini-2.5-Flash
69.8%
#8
Claude-4.0-Sonnet-Thinking
69.8%
#9
Claude-3.7-Sonnet-Thinking
69.1%
#10
OpenAI-O4-mini-medium
68.5%
#11
OpenAI-O4-mini-high
68.3%
#12
Qwen-3
68.3%
#13
OpenAI-O3-medium
68.2%
#14
OpenAI-O3-high
67.0%
#15
GPT-4o-0513
66.5%
#16
Claude-4.0-Opus
66.5%
#17
GPT-4.1
66.4%
#18
Gemini-2.5-Flash-Thinking
65.5%
#19
DeepSeek-V3
65.4%
#20
Claude-4.0-Sonnet
64.7%
#21
Grok-3-Mini-Beta
64.6%
#22
GPT-5-nano
64.4%
#23
Kimi-K2-Instruct
63.4%
#24
Mistral-Medium-3.1
60.2%
#25
GPT-4.5
60.1%
#26
Claude-3.7-Sonnet
60.0%
#27
GPT-OSS-120B
56.8%
#28
Claude-3.5-Sonnet-1022
56.2%
#29
Phi-4
54.7%
#30
Mistral-Large
54.2%
#31
Claude-3.5-Haiku
54.1%
#32
Cohere-Command-R-Plus
49.2%
#33
Cohere-Command-A
48.5%
#34
Llama-4-Maverick-17B
47.1%