Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

Chatbot Arena AAII

Evaluates model performance on the AAII (Artificial Analysis Intelligence Index), an aggregate intelligence benchmark that combines results from multiple challenging evaluations to produce a single composite score. Displayed on the Chatbot Arena leaderboard alongside Arena Elo ratings.
Source:

Model Performance

#1
Gemini 3.1 Pro Preview
76.0%
#2
Claude Opus 4.6 (Thinking)
73.0%
#3
Grok 4.20 (Reasoning)
72.0%
#4
Grok 4.20
72.0%
#5
Claude Opus 4.6
71.0%
#6
GLM-5
70.0%
#7
Claude Opus 4.5 (Thinking)
70.0%
#8
GPT-5.4
70.0%
#9
Kimi K2.5
69.0%
#10
GLM-4.7
68.0%
#11
GPT-5.2
68.0%
#12
Grok 4 (Thinking)
66.0%
#13
Grok 4
66.0%
#14
GPT-5 (Thinking)
66.0%
#15
DeepSeek V3.2 (Thinking)
66.0%
#16
OpenAI o3 (Medium Effort)
65.0%
#17
Claude Sonnet 4.6
65.0%
#18
GPT-5
65.0%
#19
Gemini 2.5 Pro (Thinking)
63.0%
#20
Gemini 2.5 Pro (Jun 2025)
63.0%
#21
OpenAI o4 Mini (High Effort)
63.0%
#22
OpenAI o4 Mini (Medium Effort)
63.0%
#23
Gemini 2.5 Pro
63.0%
#24
GPT-5 Mini (Thinking)
62.0%
#25
GPT-5 Mini
62.0%
#26
Qwen 3 (Thinking)
62.0%
#27
Qwen 3
62.0%
#28
Claude Sonnet 4.5 (Thinking)
61.0%
#29
Claude Opus 4.5
60.0%
#30
Claude 4.1 Opus (Thinking)
59.0%
#31
GPT OSS 120B
59.0%
#32
Claude 4.0 Sonnet (Thinking)
57.0%
#33
Claude 4.0 Opus (Thinking)
57.0%
#34
Grok 3 Mini
56.0%
#35
Gemini 2.5 Flash
56.0%
#36
OpenAI o3 Mini (High Effort)
53.0%
#37
GPT-5 Nano (Thinking)
52.0%
#38
DeepSeek V3.2
52.0%
#39
GPT-5 Nano
52.0%
#40
OpenAI o3 Mini (Medium Effort)
51.0%
#41
OpenAI o1
50.0%
#42
Llama 3.3 70B
50.0%
#43
Claude Sonnet 4.5
49.0%
#44
DeepSeek R1
48.0%
#45
Claude 4.1 Opus
47.0%
#46
Kimi K2
47.0%
#47
Claude 4.0 Opus
45.0%
#48
GPT-4.1
45.0%
#49
Claude 3.7 Sonnet (Thinking)
45.0%
#50
Grok 3
44.0%
#51
Llama 3.1 405B
44.0%
#52
Claude 4.0 Sonnet
44.0%
#53
Claude Haiku 4.5
42.0%
#54
DeepSeek V3
42.0%
#55
DeepSeek V3 (Mar 2025)
42.0%
#56
OpenAI o1 Mini
41.0%
#57
Llama 4 Maverick 17B
40.0%
#58
GPT-4.5
40.0%
#59
Gemini 2.0 Flash (Thinking)
40.0%
#60
GPT-4o Mini
40.0%
#61
Mistral Large 3
40.0%
#62
Gemini 2.0 Flash Thinking (Jan 2025)
40.0%
#63
Mistral Medium 3.1
37.0%
#64
Gemini 2.0 Pro
36.0%
#65
Gemini 2.0 Flash
36.0%
#66
Claude 3.7 Sonnet
35.0%
#67
Gemini 1.5 Pro
32.0%
#68
Claude 3.5 Sonnet
31.0%
#69
Cohere Command A
30.0%
#70
GPT-4o
28.0%
#71
Phi-4
26.0%
#72
Mistral Large 2
25.0%
#73
Claude 3 Opus
22.0%
#74
Claude 3.5 Haiku
21.0%
#75
Claude 3 Sonnet
14.0%
#76
GPT-3.5 Turbo
9.0%
#77
Cohere Command R+
7.0%
#78
Llama 2 70B
6.0%
#79
Llama 2 13B
6.0%