DeepTumorVQA - Measurement: leaderboard
DeepTumorVQA (2026) is a hierarchical visual question answering benchmark on abdominal 3D CT that follows the evidence chain of tumor diagnosis, scored as multiple-choice accuracy on a balanced 10,000-question test subset. This board is accuracy in percent on the four-option measurement questions on volumes, Hounsfield units and ratios, answered directly from the image input without tools; higher is better.
Metric: Accuracy (%) on the Measurement-stage questions (four numeric options: organ and lesion volume, Hounsfield units, ratios; chance 25%) of the balanced 10,000-question multiple-choice test subset of DeepTumorVQA (42 subtypes over abdominal 3D CT volumes), direct inference zero-shot with each model's native input (2D organ-centred slices for 2D models, the full 3D volume for 3D models); higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 14 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 35.7 | #175 |
| 2 | InternVL3-8B | 32 | #1782 |
| 3 | Qwen 3.5 35B A3B | 31.4 | #492 |
| 4 | Meissa-4B | 30.8 | |
| 5 | LLaVA-OneVision-7B | 30.4 | #2981 |
| 6 | Qwen 3.5 9B | 29.3 | #828 |
| 7 | Qwen 3 VL 4B | 29.1 | #1151 |
| 8 | HuatuoGPT-Vision-7B | 28.2 | #2162 |
| 9 | Qwen 3 VL 8B | 27.7 | #917 |
| 10 | M3D-LaMed-Phi-3-4B | 25 | |
| 11 | MedGemma 1.5 4B | 24.6 | #2716 |
| 12 | LLaVA-Med | 23.4 | |
| 13 | M3D-LaMed-Llama-2-7B | 23.3 | |
| 14 | RadFM | 21.7 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=deeptumorvqa-measurement · How It Works · Data refreshed daily, snapshot 2026-11-21.