DSGym - DAEval-Verified: leaderboard
Metric: Exact-Match Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K2 0905 | 92.82 |
| 2 | GPT-4o (2024-08-06) | 92.26 |
| 3 | Claude Sonnet 4.5 | 91.71 |
| 4 | Claude Sonnet 4 | 90.91 |
| 5 | GPT-5.1 (High) | 89.5 |
| 6 | GPT-5 (Medium) | 89.5 |
| 7 | GPT-5.1 (Non-reasoning) | 87.85 |
| 8 | GPT-OSS-120B | 84.53 |
| 9 | DeepSeek V3.1 | 82.32 |
| 10 | Qwen 3 4B Instruct | 64.47 |
| 11 | Qwen 2.5 7B Instruct | 50.56 |
Interactive version: theaggregate.ai/benchmark?slug=dsgym-daeval-verified · How It Works · Data refreshed daily, snapshot 2026-09-25.