Blind AI arena
Two models enter.
You crown the winner.
One prompt. Two anonymous answers. Your vote decides — and moves the live ELO leaderboard of the world's best AI models.
12models0duels0votes
Leaderboard
Benchmarks + live community votes
1
Claude 3.5 Sonnet
Anthropic · Index 78.7
78.7
Overall
2
GPT-4o
OpenAI · Index 77.9
77.9
Overall
3
Grok 2
xAI · Index 77.6
77.6
Overall
4
Qwen 2.5 72B
Qwen · Index 76.1
76.1
Overall
5
Gemini 1.5 Pro
Google · Index 71.9
71.9
Overall
6
DeepSeek Chat
DeepSeek · Index 71.6
71.6
Overall
7
Llama 3.1 70B
Meta · Index 71.2
71.2
Overall
8
GPT-4o mini
OpenAI · Index 70.5
70.5
Overall
Fair by design
01
Same prompt, both models
Two models get identical input. Zero cherry-picking.
02
Vote blind
Answers are labelled A and B. Brand bias can't creep in until you've decided.
03
ELO does the rest
Beat a stronger model, gain more points. The crowd surfaces the real winners.