healthcare benchmark
HealthBench Professional Leaderboard
HealthBench Professional evaluates model capability and safety for clinician use cases using real clinician-style chats and physician-authored grading rubrics.
Updated Oct 23, 2026
Models20
Model coverage20
MetricScore
EvidenceB
HealthBench Professional Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score69.20% | Percentile100.00% | Participants20 | EvidenceC | Evaluated |
| Rank02 | ModelAN | Score66.00% | Percentile94.74% | Participants20 | EvidenceC | Evaluated |
| Rank03 | ModelAN | Score65.60% | Percentile89.47% | Participants20 | EvidenceC | Evaluated |
| Rank04 | ModelIN | Score65.35% | Percentile84.21% | Participants20 | EvidenceC | Evaluated |
| Rank05 | ModelAN | Score64.80% | Percentile78.95% | Participants20 | EvidenceC | Evaluated |
| Rank06 | ModelOP | Score64.20% | Percentile73.68% | Participants20 | EvidenceC | Evaluated |
| Rank07 | ModelOP | Score63.40% | Percentile68.42% | Participants20 | EvidenceC | Evaluated |
| Rank08 | ModelME | Score62.30% | Percentile63.16% | Participants20 | EvidenceC | Evaluated |
| Rank09 | ModelME | Score62.30% | Percentile57.89% | Participants20 | EvidenceC | Evaluated |
| Rank10 | ModelOP | Score60.50% | Percentile52.63% | Participants20 | EvidenceC | Evaluated |
| Rank11 | ModelME | Score60.50% | Percentile47.37% | Participants20 | EvidenceC | Evaluated |
| Rank12 | ModelME | Score60.50% | Percentile42.11% | Participants20 | EvidenceC | Evaluated |
| Rank13 | ModelAN | Score59.80% | Percentile36.84% | Participants20 | EvidenceC | Evaluated |
| Rank14 | ModelAN | Score57.80% | Percentile31.58% | Participants20 | EvidenceC | Evaluated |
| Rank15 | ModelOP | Score57.70% | Percentile26.32% | Participants20 | EvidenceC | Evaluated |
| Rank16 | ModelXA | Score56.70% | Percentile21.05% | Participants20 | EvidenceC | Evaluated |
| Rank17 | ModelAN | Score55.80% | Percentile15.79% | Participants20 | EvidenceC | Evaluated |
| Rank18 | ModelOP | Score55.70% | Percentile10.53% | Participants20 | EvidenceC | Evaluated |
| Rank19 | ModelOP | Score38.40% | Percentile5.26% | Participants20 | EvidenceC | Evaluated |
| Rank20 | ModelMI | Score35.00% | Percentile0.00% | Participants20 | EvidenceC | Evaluated |
HealthBench Professional Highlights
The leading models and scores on this benchmark.
HealthBench Professional Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for HealthBench Professional
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is HealthBench Professional?
What HealthBench Professional measures and how its scores work.
HealthBench Professional is a healthcare benchmark for evaluating models in clinician use cases.
It measures model capability and safety using real clinician-style chats and physician-authored grading rubrics.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- HealthBench Professional
- Modality
- text
- Primary category
- healthcare
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about HealthBench Professional.
Which model scores highest on HealthBench Professional?
Claude Sonnet 5.5 is currently ranked first with 69.20%.
What are the top three models on HealthBench Professional?
The current leaders are Claude Sonnet 5.5 (69.20%), Claude Fable 5 (66.00%), and Claude Opus 5.5 (65.60%).
Which HealthBench Professional model has the lowest official input price?
Claude Haiku 5.5 has the lowest matched official input price at $0.10 input / $0.50 output per 1M tokens.
Which models are fastest among HealthBench Professional results?
The fastest matched records are GPT-5.6 Luna (215.52 tok/s via OpenAI), GPT-5.5 Instant (100.00 tok/s via OpenAI), and GPT-5.6 Terra (43.35 tok/s via OpenAI).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.