ClinHallu: leaderboard
Metric: Answer accuracy (%), macro-averaged over the ClinHallu test items of VQA-RAD, PathVQA, MedFrameQA and MedXpertQA (medical VQA with validated reference traces of visual recognition, knowledge recall and reasoning integration); the model writes a structured trace and answer (temperature 0.7) and a Qwen3.5-27B judge grades them (temperature 0.01); higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 80.1 |
| 2 | Qwen 3.5 9B | 69.1 |
| 3 | Qwen 3.5 4B | 64.3 |
| 4 | Qwen 3 VL 32B | 63.8 |
| 5 | InternVL3.5-8B | 53.9 |
| 6 | MedGemma-4B-IT | 53.2 |
| 7 | Lingshu-7B | 52.7 |
| 8 | Qwen 2.5 VL 7B Instruct | 42.7 |
Interactive version: theaggregate.ai/benchmark?slug=clinhallu · How It Works · Data refreshed daily, snapshot 2026-09-29.