WuYuEval - Foundation Module - Accuracy: leaderboard
Accuracy on the Foundation Module of WuYuEval (arXiv 2608.07529v1, Appendix Table 7): 4,590 audited closed-ended multiple-choice questions on solid waste management across six task types and eight domains. Percent; higher is better. Standard and thinking modes of the same model are separate rows.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 94.64 | #87 |
| 2 | Gemini 3 Pro (Preview) | 92.77 | #71 |
| 3 | GLM-4.6 (Non-reasoning) | 92.31 | #265 (GLM-4.6) |
| 4 | DeepSeek V3.2 (Thinking) | 92.16 | #210 (DeepSeek V3.2) |
| 5 | Qwen 3 Max | 91.83 | #213 |
| 6 | DeepSeek V3.2 (Non-reasoning) | 91.37 | #210 (DeepSeek V3.2) |
| 7 | Kimi K2 (Thinking) | 91.26 | #252 (Kimi K2) |
| 8 | Qwen 3 235B A22B (Thinking) | 90.83 | #323 (Qwen 3 235B A22B) |
| 9 | GPT-5.2 | 90.04 | #115 |
| 10 | Qwen 3 235B A22B Instruct | 89.63 | #336 |
| 11 | Qwen 3 32B (Thinking) | 87.86 | #449 (Qwen 3 32B) |
| 12 | Qwen 3 32B (Non-reasoning) | 87.49 | #449 (Qwen 3 32B) |
| 13 | GLM-4.6 (Thinking) | 84.97 | #265 (GLM-4.6) |
| 14 | Gemma 3 27B (IT) | 82.29 | #528 |
| 15 | Qwen 3 14B (Reasoning) | 81.37 | #542 (Qwen 3 14B) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=wuyueval-foundation-module-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-05.