Arena-Hard v2 (GPT-4.1 Judge) — leaderboard

500 challenging real-world queries auto-evaluated by GPT-4.1 as judge, with style control. Alternative judging config for Arena-Hard v2.

Metric: Win Rate (%). Source: github.com. Status: saturation imminent. 28 models tracked.

Top models

#ModelScore
1O3 (2025-04-16)87
2O4 Mini (2025-04-16) (High)81.7
3O4 Mini (2025-04-16)78
4O3 Mini (2025-01-31) (High)64.8
5O1 (2024-12-17) (High)58.7
6GPT-4.158.3
7O1 (2024-12-17)50.2
8O3 Mini (2025-01-31)50
9GPT-4.1 Mini48.6
10DeepSeek R148
11Qwen 3 235B A22B46.7
12Gemini 2.5 Flash45.1
13GPT-4.5 (Preview)43
14QwQ-32B36.1
15Qwen 3 32B35.8

Interactive version: theaggregate.ai/benchmark?slug=arena-hard-v2-gpt-4-1-judge · How the rankings work · Data refreshed daily, snapshot 2026-07-22.