Patient-Context Ambiguity: leaderboard
Dietary-safety classification accuracy on 487 queries after the Agent framework elicits missing patient context. Basic direct prompting and RaR rephrasing are separate configurations; this is not diagnosis-retrieval accuracy.
Metric: Accuracy (%, Agent dietary-safety classification). Source: arxiv.org. Saturation forecast: Estimated already saturated. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 | 91.8 | #37 |
| 2 | Gemini 3.1 Pro (Preview) | 90.3 | #58 |
| 3 | GPT-5.5 | 89.3 | #31 |
| 4 | Llama 3.3 70B | 86.9 | #579 |
| 5 | Mistral Large | 86.4 | #710 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=patient-context-ambiguity · How It Works · Data refreshed daily, snapshot 2026-10-20.