ADRD-Bench - Unified QA: leaderboard
Metric: Exact-match accuracy (0-1, shown times 100) on the 1,438 Alzheimer's-disease and dementia-related items consolidated from MedMCQA, MedQA, HEAD-QA, MEDEC, MedBullets, MedHallu and PubMedQA (multiple-choice and medical error detection), pooled over all items; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 36 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.1 70B Instruct | 92.98 | #548 |
| 2 | GPT-5.5 (2026-04-23) | 92.77 | #21 |
| 3 | Claude Opus 4.7 | 92.21 | #45 |
| 4 | Gemini 3.1 Pro (Preview) | 91.66 | #54 |
| 5 | Claude Opus 4.5 (20251101) | 91.1 | #70 |
| 6 | GPT-5.2 | 89.43 | #105 |
| 7 | Qwen 3 Max (2025-09-23) | 89.08 | #249 |
| 8 | Qwen 3 235B A22B 2507 Instruct | 87.69 | #291 |
| 9 | Llama 3 8B Instruct | 87.27 | #1115 |
| 10 | Llama 3.1 8B Instruct | 83.31 | #1018 |
| 11 | Qwen 2.5 72B Instruct | 83.03 | #436 |
| 12 | Qwen 3 30B A3B 2507 Instruct | 82.68 | #464 |
| 13 | Grok 4.1 Fast (Non-reasoning) | 82.06 | #208 (Grok 4.1 Fast) |
| 14 | Qwen 2.5 14B Instruct | 75.17 | #634 |
| 15 | Phi-3-medium-4k-instruct | 74.76 | #939 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=adrd-bench-unified-qa · How It Works · Data refreshed daily, snapshot 2026-10-11.