DeepTumorVQA: leaderboard
DeepTumorVQA evaluates medical vision-language models and agents on 3D CT reasoning across recognition, measurement, visual reasoning, and medical reasoning stages. The benchmark is image-based, with tool-interaction support, and the pinned Overall Accuracy (%) measures end-to-end answer correctness. Source: BenchmarkList’s attributed result table.
Metric: Overall Accuracy (%). Source: benchmarklist.com. Saturation forecast: Around 2029. 29 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 35B A3B | 39.9 | #268 |
| 2 | Gemini 3 Flash | 38.8 | #103 |
| 3 | Qwen 3.5 9B | 38.5 | #391 |
| 4 | InternVL3-8B | 34.4 | #624 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=deeptumorvqa · How It Works · Data refreshed daily, snapshot 2026-11-05.