Arena-Hard v2

Arena-Hard v2 - instruction-following chat benchmark; win rate evaluated by GPT-4.1 for reproducibility.

Category

Instruction Following

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for Arena-Hard v2

Model Score Date Verified
Qwen3-Next-80B-A3B-Thinking
100.00%
10.09.2025 Verified
Qwen3-VL-30B-A3B
58.50
04.10.2025 Verified