WuYuEval - Foundation Module - Macro-F1: leaderboard
Metric: Macro-F1 (%). Source: arxiv.org. 33 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 | 95.73 |
| 2 | Gemini 3 Pro (Preview) | 95.53 |
| 3 | GLM-4.6 (Non-reasoning) | 93.54 |
| 4 | DeepSeek V3.2 (Thinking) | 93.51 |
| 5 | Qwen 3 Max | 93.09 |
| 6 | DeepSeek V3.2 (Non-reasoning) | 92.66 |
| 7 | Qwen 3 235B A22B (Thinking) | 92.54 |
| 8 | Kimi K2 (Thinking) | 92.23 |
| 9 | GPT-5.2 | 91.36 |
| 10 | Qwen 3 235B A22B Instruct | 90.93 |
| 11 | Qwen 3 32B (Thinking) | 89.58 |
| 12 | Qwen 3 32B (Non-reasoning) | 89.18 |
| 13 | Qwen 3 14B (Reasoning) | 87.7 |
| 14 | GLM-4.6 (Thinking) | 86.53 |
| 15 | Gemma 3 27B (IT) | 84.73 |
Interactive version: theaggregate.ai/benchmark?slug=wuyueval-foundation-module-macro-f1 · How It Works · Data refreshed daily, snapshot 2026-09-19.