Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

LiveBench (Reasoning)

Evaluates performance on LiveBench reasoning tasks, including harder versions of Big-Bench Hard's Web of Lies and Zebra Puzzles.
Source:

Model Performance

#1
GPT-5
98.2%
#2
Claude-4.0-Sonnet-Thinking
95.2%
#3
OpenAI-O3-high
94.7%
#4
Gemini-2.5-Pro-0325
93.7%
#5
Claude-4.1-Opus-Thinking
93.2%
#6
DeepSeek-R1
91.1%
#7
OpenAI-O3-medium
91.0%
#8
Claude-4.0-Opus-Thinking
90.5%
#9
OpenAI-O4-mini-high
88.1%
#10
Grok-3-Mini-Beta
87.6%
#11
Qwen-3
83.1%
#12
GPT-5-mini
82.6%
#13
Gemini-2.5-Flash
78.5%
#14
OpenAI-O4-mini-medium
78.5%
#15
GPT-OSS-120B
77.6%
#16
Claude-3.7-Sonnet-Thinking
76.2%
#17
Gemini-2.5-Flash-Thinking
63.4%
#18
Kimi-K2-Instruct
63.0%
#19
DeepSeek-V3
59.2%
#20
Phi-4
57.8%
#21
Claude-4.0-Opus
56.4%
#22
GPT-5-nano
54.9%
#23
Claude-4.0-Sonnet
54.9%
#24
GPT-4.5
54.4%
#25
Claude-3.7-Sonnet
49.1%
#26
GPT-4o-0513
48.8%
#27
GPT-4.1
44.4%
#28
Llama-4-Maverick-17B
43.8%
#29
Claude-3.5-Sonnet-1022
43.2%
#30
Mistral-Medium-3.1
42.0%
#31
Cohere-Command-A
36.3%
#32
Mistral-Large
33.8%
#33
Claude-3.5-Haiku
26.2%
#34
Cohere-Command-R-Plus
21.6%