leaderboard.cn

Explore

Frontier Models
Historical Trends
Local Models
Compare Models
Ranking Studio

Library

Models
Pricing
Organizations
Benchmarks
Data sources

leaderboard.cn

Benchmark details loading

Loading page content

About Ranking methodology Changelog

v0.1.26.a85cb1563d03浙ICP备09076226号-7

Back to benchmarks

Arena-Hard v2Conversation

Results for 9 models collected from 2 public sources.

#ModelScore

Qwen3-235B-A22B-Thinking-2507Alibaba

79.7%

Qwen3-235B-A22B-Instruct-2507Alibaba

79.2%

Qwen3-30B-A3B-Instruct-2507Alibaba

69.0%

Qwen3-30B-A3B-Thinking-2507Alibaba

56.0%

Qwen3-4B-Instruct-2507Alibaba

43.4%

Qwen3-4B-Thinking-2507Alibaba

34.9%

Qwen3-30B-A3BAlibaba · Initial April 2025 model · non-reasoning mode

24.8%

GPT-4.1 nano 2025-04-14OpenAI

15.9%

Qwen3-4BAlibaba · Initial April 2025 model · non-reasoning mode

9.5%

9 models total