AbstentionBench - underspecified context - Musique - Precision: leaderboard
Metric: Abstention Precision (%). Source: github.com. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.1 70B | 99.41 |
| 2 | Llama 3.1 8B | 99.31 |
| 3 | Llama-3.1-Tulu-3-8B | 98.73 |
| 4 | Llama-3.1-Tulu-3-8B-SFT | 97.06 |
| 5 | Llama-3.1-Tulu-3-8B-DPO | 95.31 |
| 6 | GPT-4o | 91.68 |
| 7 | O1 | 91.29 |
| 8 | Llama 3.1 70B Instruct | 90.11 |
| 9 | Llama 3.1 405B Instruct | 89.41 |
| 10 | Llama-3.1-Tulu-3-70B-DPO | 89.24 |
| 11 | Llama-3.1-Tulu-3-70B | 88.97 |
| 12 | DeepSeek R1 Distill Llama 70B | 88.66 |
| 13 | Llama 3.3 70B Instruct | 88.42 |
| 14 | Llama 3.1 8B Instruct | 83.98 |
| 15 | Gemini 1.5 Pro | 80.99 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-musique-precision · How It Works · Data refreshed daily, snapshot 2026-09-19.