healthcare benchmark
HealthBench Consensus
HealthBench Consensus is a HealthBench subset focused on questions where physician-created rubric criteria have especially high agreement, measuring healthcare performance and safety on consensus-evaluable conversations.
Updated Aug 7, 2026
Published models4
Registry coverage4
MetricScore
EvidenceB
HealthBench Consensus leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | OP | 95.5% | 100.0% | 4 | C | |
| 2 | OP | 95.1% | 66.7% | 4 | C | |
| 3 | OP | 95.1% | 33.3% | 4 | C | |
| 4 | OP | 94.7% | 0.0% | 4 | C |
Score distribution
Top published rows on the benchmark's original scale.
HealthBench Consensus highlights
The top published results on this benchmark's own scale.
What is HealthBench Consensus?
Definition and scoring fields from the benchmark registry.
HealthBench Consensus is a HealthBench subset focused on questions where physician-created rubric criteria have especially high agreement, measuring healthcare performance and safety on consensus-evaluable conversations.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- HealthBench Consensus
- Modality
- text
- Primary category
- healthcare
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- healthbench-consensus|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about HealthBench Consensus.
Which model scores highest on HealthBench Consensus?
GPT-5.6 Sol is currently ranked first with 95.5%.
What does HealthBench Consensus measure?
HealthBench Consensus is a HealthBench subset focused on questions where physician-created rubric criteria have especially high agreement, measuring healthcare performance and safety on consensus-evaluable conversations.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
4 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.