the ai forger
Models
Organizations
Providers
Benchmarks
Compare

Arena Hard

general
text
About

Arena Hard benchmark

Evaluation Stats
Total Models22
Organizations7
Verified Results0
Self-Reported22
Benchmark Details
Max Score1
Language
en
Performance Overview
Score distribution and top performers

Score Distribution

22 models
Top Score
95.6%
Average Score
66.4%
High Performers (80%+)
7

Top Organizations

#1NVIDIA
1 model
88.3%
#2DeepSeek
2 models
84.3%
#3Alibaba
5 models
82.7%
#4Mistral AI
3 models
67.2%
#5Microsoft
6 models
55.9%
Leaderboard
Top 20 models ranked by performance
1
95.6%
Raw: 0.956
Self-reported
2
93.8%
Raw: 0.938
Self-reported
3
92.3%
Raw: 0.923
Self-reported
4
91.0%
Raw: 0.91
Self-reported
5
88.3%
Raw: 0.883
Self-reported
6
87.6%
Raw: 0.876
Self-reported
7
81.2%
Raw: 0.812
Self-reported
8
79.0%
Raw: 0.79
Self-reported
9
76.2%
Raw: 0.762
Self-reported
10
75.4%
Raw: 0.754
Self-reported
11
73.3%
Raw: 0.733
Self-reported
12
70.9%
Raw: 0.709
Self-reported
13
65.4%
Raw: 0.654
Self-reported
14
57.6%
Raw: 0.5756
Self-reported
15
57.6%
Raw: 0.5756
Self-reported
16
52.0%
Raw: 0.52
Self-reported
17
46.1%
Raw: 0.461
Self-reported
18
43.1%
Raw: 0.431
Self-reported
19
37.9%
Raw: 0.379
Self-reported
20
37.0%
Raw: 0.37
Self-reported
Showing top 20 of 22 models