TransClean Generation - Prompt 0 - Explanatory Text: leaderboard
Metric: Explanatory Noise Rate (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 30B A3B 2507 (Thinking) | 2.81 |
| 2 | Qwen 3 4B 2507 (Thinking) | 2.84 |
| 3 | Qwen 3 30B A3B 2507 Instruct | 2.95 |
| 4 | Qwen 3 4B 2507 Instruct | 3.03 |
| 5 | aya-expanse-32B | 15.03 |
| 6 | DeepSeek R1 Distill Qwen 32B | 22.22 |
| 7 | Llama 3.2 3B Instruct | 29.47 |
| 8 | Qwen 2.5 32B Instruct | 53.93 |
| 9 | Gemma 3 27B (IT) | 97.32 |
Interactive version: theaggregate.ai/benchmark?slug=transclean-generation-prompt-0-explanatory-text · How It Works · Data refreshed daily, snapshot 2026-09-24.