AbstentionBench - underspecified context - Musique - Recall: leaderboard
Metric: Abstention Recall (%). Source: github.com. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 32B Instruct | 87.08 |
| 2 | Mistral 7B Instruct (v0.3) | 86.4 |
| 3 | Gemini 1.5 Pro | 83.51 |
| 4 | Llama 3.1 8B Instruct | 80.85 |
| 5 | Llama 3.3 70B Instruct | 78.34 |
| 6 | Llama 3.1 405B Instruct | 76.98 |
| 7 | Llama 3.1 70B Instruct | 73.4 |
| 8 | GPT-4o | 62.77 |
| 9 | Llama 3.1 8B | 54.71 |
| 10 | Llama-3.1-Tulu-3-70B-DPO | 54.18 |
| 11 | Llama 3.1 70B | 51.52 |
| 12 | O1 | 48.56 |
| 13 | Llama-3.1-Tulu-3-70B | 47.19 |
| 14 | DeepSeek R1 Distill Llama 70B | 36.25 |
| 15 | Llama-3.1-Tulu-3-8B-DPO | 18.54 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-musique-recall · How It Works · Data refreshed daily, snapshot 2026-09-19.