# Healthcare benchmarks

Find tests for health conversations, clinical documentation, billing, and medical questions.

Canonical page: https://benchlm.ai/benchmarks/collections/healthcare

10 records. Conversation rubrics, medical exams, and administrative tasks measure different work. Their scores do not establish clinical safety or patient outcomes.

Records keep metrics and harness variants separate. Source rows can include different model and agent configurations; linked profiles count distinct public model identities. Public model rows link to original published artifacts, which may include publisher comparisons of other providers' models. These counts do not represent independent evaluations. Dates describe the stored source record.

## Health and clinician conversations

| Benchmark | Stored coverage | Task and run conditions | Original source |
|---|---|---|---|
| [HealthBench (raw)](/benchmarks/healthbench) | 7 public model rows linked to original artifacts | The raw HealthBench score stays separate from the length-adjusted score. 5,000 multi-turn patient conversations | [Original source](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |
| [HealthBench (length-adjusted)](/benchmarks/healthbenchlengthadjusted) | 7 public model rows linked to original artifacts | A length adjustment changes the metric; inspect it beside the raw score. 5,000 multi-turn patient conversations | [Original source](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |
| [HealthBench Hard](/benchmarks/healthbench-hard) | 12 public model rows linked to original artifacts | The harder subset is a separate evaluation. Inspect row notes for raw or length-adjusted conditions. 1,000 health prompts | [Original source](https://ai.meta.com/static-resource/muse-spark-eval-methodology) |
| [HealthBench Professional](/benchmarks/healthbenchprofessional) | 13 public model rows linked to original artifacts | Clinician-facing tasks can include product harnesses; read the run conditions. Clinician chat tasks | [Original source](https://arxiv.org/abs/2604.27470) |
| [HealthBench Professional (raw)](/benchmarks/healthbenchprofessionalraw) | 7 public model rows linked to original artifacts | The raw Professional score stays separate from the length-adjusted result. 525 physician-authored conversations | [Original source](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |

## Documentation and billing

| Benchmark | Stored coverage | Task and run conditions | Original source |
|---|---|---|---|
| [MedScribe](/benchmarks/valsmedscribe) | 109 stored source rows; 80 linked model profiles | SOAP-note documentation is evaluated from transcripts; it does not measure an ambient-scribing product end to end. Medical administrative support tasks Source update in snapshot: 2026-10-13; Source snapshot captured: 2026-10-15T18:47:28.403Z | [Original source](https://www.vals.ai/benchmarks/medscribe) |
| [MedCode](/benchmarks/valsmedcode) | 107 stored source rows; 78 linked model profiles | Medical billing support is administrative evidence, separate from diagnosis or treatment. Medical billing support tasks Source update in snapshot: 2026-10-13; Source snapshot captured: 2026-10-15T18:47:28.403Z | [Original source](https://www.vals.ai/benchmarks/medcode) |

## Medical questions

| Benchmark | Stored coverage | Task and run conditions | Original source |
|---|---|---|---|
| [MedXpertQA (Text)](/benchmarks/medxpertqatext) | 5 public model rows linked to original artifacts | Text-only medical multiple-choice results remain separate from the multimodal test. 2,450 medical multiple-choice questions | [Original source](https://ai.meta.com/static-resource/muse-spark-eval-methodology) |
| [MedXpertQA (MM)](/benchmarks/medxpertqamm) | 8 public model rows linked to original artifacts | Medical-image multiple-choice results use their own task set. 2,000 multimodal medical questions | [Original source](https://ai.meta.com/static-resource/muse-spark-eval-methodology) |

## Historical evidence

| Benchmark | Stored coverage | Task and run conditions | Original source |
|---|---|---|---|
| [MedQA](/benchmarks/valsmedqa) | 95 stored source rows; 39 linked model profiles | Archived source board. Vals archived this medical-question board after saturation; it is not a current-release test. Medical question answering Source update in snapshot: 2026-04-16; Source snapshot captured: 2026-10-15T18:47:28.403Z | [Original source](https://www.vals.ai/benchmarks/medqa) |

## Other collections

- [Finance benchmarks](/benchmarks/collections/finance)
- [Legal benchmarks](/benchmarks/collections/legal)

[All benchmarks](/benchmarks)