AbstentionBench - underspecified context - MediQ - Precision: leaderboard
Metric: Abstention Precision (%). Source: github.com. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama-3.1-Tulu-3-8B | 100 |
| 2 | GPT-4o | 99.39 |
| 3 | Llama 3.1 405B Instruct | 99.35 |
| 4 | Llama 3.1 70B Instruct | 99.22 |
| 5 | Llama 3.1 70B | 99.09 |
| 6 | Llama-3.1-Tulu-3-70B | 99.07 |
| 7 | Llama 3.1 8B | 98.92 |
| 8 | Gemini 1.5 Pro | 98.62 |
| 9 | O1 | 98.53 |
| 10 | Llama 3.3 70B Instruct | 98.19 |
| 11 | Llama-3.1-Tulu-3-70B-DPO | 97.79 |
| 12 | Llama 3.1 8B Instruct | 97.41 |
| 13 | DeepSeek R1 Distill Llama 70B | 96.23 |
| 14 | Mistral 7B Instruct (v0.3) | 95.58 |
| 15 | Llama-3.1-Tulu-3-8B-SFT | 95.45 |
Interactive version: theaggregate.ai/benchmark?slug=abstentionbench-underspecified-context-mediq-precision · How It Works · Data refreshed daily, snapshot 2026-09-19.