AbstentionBench - underspecified context - GSM8K - Recall: leaderboard
Metric: Abstention Recall (%). Source: github.com. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O1 (Low) | 95.63 |
| 2 | O1 (High) | 95.47 |
| 3 | O1 | 95.38 |
| 4 | Llama 3.1 405B Instruct | 95.14 |
| 5 | Qwen 2.5 32B Instruct | 94.48 |
| 6 | Llama 3.1 70B Instruct | 94.31 |
| 7 | Gemini 1.5 Pro | 94.23 |
| 8 | Llama 3.1 8B Instruct | 92.66 |
| 9 | GPT-4o | 92 |
| 10 | Llama 3.3 70B Instruct | 91.1 |
| 11 | Llama-3.1-Tulu-3-70B-DPO | 89.12 |
| 12 | Llama-3.1-Tulu-3-70B | 87.8 |
| 13 | Mistral 7B Instruct (v0.3) | 87.06 |
| 14 | Llama-3.1-Tulu-3-8B-DPO | 75.35 |
| 15 | Llama-3.1-Tulu-3-8B | 74.53 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-gsm8k-recall · How It Works · Data refreshed daily, snapshot 2026-09-19.