DrVD-Bench: leaderboard
DrVD-Bench evaluates multimodal clinical visual reasoning for medical image diagnosis, including evidence comprehension, reasoning trajectory assessment, and report generation. It is image-based across CT, MRI, ultrasound, radiography, and pathology, and the pinned Overall Accuracy (%) measures overall correctness. Source: BenchmarkList’s attributed result table.
Metric: Overall Accuracy (%). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 69.3 | #159 |
| 2 | O3 | 65.8 | #132 |
| 3 | GPT-4o | 64.6 | #357 |
| 4 | Claude 3.7 Sonnet | 63.4 | #259 |
| 5 | Qwen-VL-Max | 62.8 | #475 |
| 6 | Grok 3 | 60.4 | #313 |
| 7 | Qwen 2.5 VL 7B | 54.2 | #641 |
| 8 | phi-4-14B | 53.4 | #749 |
| 9 | O1 | 51.7 | #190 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=drvd-bench · How It Works · Data refreshed daily, snapshot 2026-11-05.