AbstentionBench - underspecified context - BBQ - Precision: leaderboard
Metric: Abstention Precision (%). Source: github.com. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek R1 Distill Llama 70B | 98.59 |
| 2 | Llama 3.3 70B Instruct | 98 |
| 3 | O1 | 97.6 |
| 4 | Llama 3.1 70B Instruct | 96.58 |
| 5 | Qwen 2.5 32B Instruct | 96.52 |
| 6 | Llama 3.1 405B Instruct | 96.5 |
| 7 | GPT-4o | 94.07 |
| 8 | Llama-3.1-Tulu-3-70B-DPO | 94.04 |
| 9 | Llama-3.1-Tulu-3-70B | 93.77 |
| 10 | Llama 3.1 70B | 93.28 |
| 11 | Llama-3.1-Tulu-3-8B | 93.24 |
| 12 | Llama 3.1 8B | 91.95 |
| 13 | Gemini 1.5 Pro | 91.89 |
| 14 | Llama-3.1-Tulu-3-8B-DPO | 91.41 |
| 15 | Mistral 7B Instruct (v0.3) | 86.26 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-bbq-precision · How It Works · Data refreshed daily, snapshot 2026-09-19.