Oncology Report-Derived VQA - Liver CT Report-Derived: leaderboard
Zero-shot multiple-choice visual question answering on liver CT of confirmed cancer patients from a private in-house cohort, with questions written by an LLM from each radiology report's findings and impression and checked against the report (Liu et al., 2026). The model sees one representative slice with the question and options, never the report. Accuracy in percent over valid responses; higher is better.
Metric: Accuracy (%; zero-shot multiple-choice VQA with 2-5 options on one representative 2D slice of a 3D study, the model sees the image, question and options but never the report; invalid responses excluded; liver CT, 1,000 cases, 8,901 LLM-written questions checked against the report). Source: arxiv.org. Saturation forecast: Around 2031. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 61.7 | #100 |
| 2 | MedGemma 1.5 4B | 47.8 | #2579 |
| 3 | GPT-5.2 | 44 | #172 |
| 4 | GPT-5 | 40.6 | #163 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=oncology-report-derived-vqa-liver-ct-report-derived · How It Works · Data refreshed daily, snapshot 2026-11-12.