AbstentionBench - underspecified context - UMWP - Recall: leaderboard
Metric: Abstention Recall (%). Source: github.com. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O1 | 76.11 |
| 2 | Llama 3.1 70B Instruct | 70.63 |
| 3 | Gemini 1.5 Pro | 68.23 |
| 4 | Qwen 2.5 32B Instruct | 66.23 |
| 5 | GPT-4o | 66.17 |
| 6 | Llama 3.1 8B Instruct | 61.66 |
| 7 | Mistral 7B Instruct (v0.3) | 60.63 |
| 8 | DeepSeek R1 Distill Llama 70B | 56.69 |
| 9 | Llama 3.1 70B | 54.15 |
| 10 | Llama 3.3 70B Instruct | 53.37 |
| 11 | Llama 3.1 8B | 50.6 |
| 12 | Llama-3.1-Tulu-3-70B | 49.83 |
| 13 | Llama-3.1-Tulu-3-70B-DPO | 49.14 |
| 14 | Llama-3.1-Tulu-3-8B-DPO | 31.77 |
| 15 | Llama-3.1-Tulu-3-8B | 31.6 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-umwp-recall · How It Works · Data refreshed daily, snapshot 2026-09-19.