AbstentionBench - underspecified context - BBQ - Recall: leaderboard
Metric: Abstention Recall (%). Source: github.com. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 32B Instruct | 95.05 |
| 2 | Gemini 1.5 Pro | 92.41 |
| 3 | Mistral 7B Instruct (v0.3) | 90.78 |
| 4 | GPT-4o | 90.04 |
| 5 | Llama 3.1 8B Instruct | 89.93 |
| 6 | Llama 3.1 405B Instruct | 89.82 |
| 7 | Llama-3.1-Tulu-3-70B | 88.08 |
| 8 | Llama-3.1-Tulu-3-70B-DPO | 87.85 |
| 9 | Llama 3.1 70B Instruct | 84.2 |
| 10 | Llama 3.3 70B Instruct | 79.75 |
| 11 | Llama-3.1-Tulu-3-8B-DPO | 65.86 |
| 12 | DeepSeek R1 Distill Llama 70B | 63.05 |
| 13 | Llama-3.1-Tulu-3-8B | 62.09 |
| 14 | O1 | 54.78 |
| 15 | Llama 3.1 70B | 46.06 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-bbq-recall · How It Works · Data refreshed daily, snapshot 2026-09-19.