ClinHallu
Models answer medical image questions and produce intermediate reasoning. The evaluation reports answer accuracy and separate hallucination rates for visual recognition, knowledge recall, and reasoning integration against annotated reasoning traces.
The suite adds reasoning annotations and stage-level scoring to questions from four existing medical benchmarks. Accuracy and the three hallucination rates have different directions and must remain separate.
- Made by
- Alibaba DAMO Academy
- Released
- June 2026
- Kind
- Medicine
- Level
- Expert
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Subject
- Clinical medicine
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Vision, Biology and medicine
- Strongest model tested
-
Gemini 3 Flash #60 on the capability index
- Last model added
- February 2026
- Scores refreshed
- November 18th, 2026
In the authors’ words
Building trustworthy medical multimodal large language models (MLLMs) is critical for reliable clinical decision support. Existing medical hallucination benchmarks mainly focus on data collection, but often ignore where hallucinations originate within the reasoning process. We find that hallucination sources vary across samples: errors may arise from visual misrecognition, incorrect medical knowledge recall, or flawed reasoning integration. To enable source-level hallucination diagnosis, we introduce ClinHallu, a benchmark for stage-wise hallucination diagnosis in medical MLLM reasoning. ClinHallu contains 7,031 validated instances, where each instance is augmented with a structured reasoning trace decomposed into Visual Recognition, Knowledge Recall, and Reasoning Integration. We also use stage-replacement interventions to measure how correcting specific stages affects the final answer. Beyond evaluation, we show that trace-supervised fine-tuning reduces stage-wise hallucinations. ClinHallu provides a fine-grained hallucination testbed for diagnosing and mitigating reasoning failures in medical MLLMs. The benchmark is publicly available at https://github.com/alibaba-damo-academy/ClinHallu.
Gemini 3 Flash leads with 80.1% among models released by February 2026 Gemini 3 Flash leads with 80.1% among models released by February 2026
Best score per entry from the benchmark's own leaderboard, read on November 18th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in February 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1
Gemini 3 Flash 80.1%
- 2
Qwen3.5-9B 69.1%
- 3
Qwen3-VL-Plus 67.2%
- 4
Qwen3.5-4B 64.3%
- 5
Qwen3-VL-32B-Instruct 63.8%
- 6
Qwen3-VL-Flash 63.6%
- 7
InternVL3.5-8B 53.9%
- 8 MedGemma-4B 53.2%
- 9 Lingshu-7B 52.7%
- 10
Qwen3-VL-8B-Instruct 51.5%
- 11
Qwen2.5-VL-7B 42.7%
Source: Official ClinHallu leaderboard
- 1
Gemini 3 Flash 80.1%
- 2
Qwen3.5-9B 69.1%
- 3
Qwen3-VL-Plus 67.2%
- 4
Qwen3.5-4B 64.3%
- 5
Qwen3-VL-32B-Instruct 63.8%
- 6
Qwen3-VL-Flash 63.6%
- 7
InternVL3.5-8B 53.9%
- 8
Qwen3-VL-8B-Instruct 51.5%
- 9
Qwen2.5-VL-7B 42.7%
Source: Official ClinHallu leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 80.1% | Dec 2025 | Official leaderboard | |
| 2 | | 69.1% | Feb 2026 | Official leaderboard | |
| 3 | | 67.2% | Sep 2025 | Official leaderboard | |
| 4 | | 64.3% | Feb 2026 | Official leaderboard | |
| 5 | | 63.8% | Oct 2025 | Official leaderboard | |
| 6 | | 63.6% | Oct 2025 | Official leaderboard | |
| 7 | | 53.9% | Aug 2025 | Official leaderboard | |
| 8 | MedGemma-4B | 53.2% | Official leaderboard | ||
| 9 | Lingshu-7B | 52.7% | Official leaderboard | ||
| 10 | | 51.5% | Oct 2025 | Official leaderboard | |
| 11 | | 42.7% | Jan 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 80.1% | Dec 2025 | Official leaderboard | |
| 2 | | 69.1% | Feb 2026 | Official leaderboard | |
| 3 | | 67.2% | Sep 2025 | Official leaderboard | |
| 4 | | 64.3% | Feb 2026 | Official leaderboard | |
| 5 | | 63.8% | Oct 2025 | Official leaderboard | |
| 6 | | 63.6% | Oct 2025 | Official leaderboard | |
| 7 | | 53.9% | Aug 2025 | Official leaderboard | |
| 8 | | 51.5% | Oct 2025 | Official leaderboard | |
| 9 | | 42.7% | Jan 2025 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official ClinHallu leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official ClinHallu leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Law, finance and medicine
- MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy.
Qwen 3.8 Max 80.4%
- HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
- HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
- FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy.
Ling-3.1-flash 81.7%
- KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. o1-preview 92.7%
- Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 9 months ago, in February 2026, for Qwen3.5-4B. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the official leaderboard (11 entries).
- Pass Models scored 9 models we track, plus 2 other entries on its leaderboard.
- Warning Newest score The newest came 9 months ago, in February 2026, for Qwen3.5-4B.
- Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3 Flash (#60).
- Pass Headroom The best is 80.1%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on github.com: 9 rows for models we track and 2 other entries at the last read.
- Pass Last read November 18th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.