ClinHallu - Reasoning Integration Hallucination: leaderboard

Metric: Reasoning integration hallucination rate (%): share of items whose reasoning stage, generated after the reference visual and knowledge stages are supplied, the judge marks as hallucinated, 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); lower is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.

Top models

#ModelScore
1Gemini 3 Flash2.3
2Qwen 3 VL 32B4.4
3Qwen 3.5 9B4.8
4Qwen 3.5 4B5.1
5InternVL3.5-8B6.6
6Lingshu-7B13.6
7Qwen 2.5 VL 7B Instruct18.1
8MedGemma-4B-IT30.5

Interactive version: theaggregate.ai/benchmark?slug=clinhallu-reasoning-integration-hallucination · How It Works · Data refreshed daily, snapshot 2026-09-29.