AbstentionBench - underspecified context - GPQA-Diamond - Recall: leaderboard
Metric: Abstention Recall (%). Source: github.com. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O1 | 77.5 |
| 2 | GPT-4o | 75 |
| 3 | Llama 3.1 405B Instruct | 65 |
| 4 | Qwen 2.5 32B Instruct | 65 |
| 5 | Llama-3.1-Tulu-3-70B | 65 |
| 6 | Llama-3.1-Tulu-3-70B-DPO | 65 |
| 7 | O1 (Low) | 62.5 |
| 8 | O1 (High) | 60 |
| 9 | Llama 3.1 8B Instruct | 52.5 |
| 10 | Mistral 7B Instruct (v0.3) | 52.5 |
| 11 | Gemini 1.5 Pro | 50 |
| 12 | Llama 3.1 70B Instruct | 45 |
| 13 | Llama 3.3 70B Instruct | 37.5 |
| 14 | Llama-3.1-Tulu-3-8B-SFT | 32.5 |
| 15 | DeepSeek R1 Distill Llama 70B | 25 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-gpqa-diamond-recall · How It Works · Data refreshed daily, snapshot 2026-09-19.