# HealthBench Professional Guide

> Independent HealthBench Professional analysis: task selection, clinician workflows, system harness comparisons and an interactive length-adjustment calculator.

HealthBench Professional combines physician-authored tasks, deliberate difficulty selection and a length-adjusted rubric score. Those choices make its number useful for a particular kind of comparison. This publication explains the sampling, shows historical same-model system results and makes the adjustment formula interactive. Read the benchmark dossier, inspect the calculator and use the guides to separate measured performance from broader clinical claims. Arcophos publishes independent analysis; OpenAI and the original research team created the benchmark.

## Provenance
Independent analysis published by Arcophos. Benchmark creation belongs to the credited authors. Result rows are selected paper-reported measurements with their source versions and evaluation conditions, not new Arcophos runs or a live leaderboard.

## Benchmark dossiers
- [HealthBench Professional](https://healthbenchprofessional.ai/benchmarks/healthbench-professional/): Professional task scores depend on sampling, response length and the harness. Source version: April 2026 paper v1; length-adjusted primary score.

## Original analyses
- [How HealthBench Professional adjusts for answer length](https://healthbenchprofessional.ai/guides/healthbench-professional-length-adjustment/): Apply the published coefficient in the correct units and keep example-level values separate from aggregate clipping.
- [Why 525 selected tasks do not estimate routine clinical accuracy](https://healthbenchprofessional.ai/guides/healthbench-professional-sampling-and-routine-use/): Read difficulty enrichment and use-case composition before generalizing a HealthBench Professional score.
- [Model versus harness: reading the GPT-5.4 comparison](https://healthbenchprofessional.ai/guides/healthbench-professional-model-versus-harness/): Interpret the original same-model system experiment without attributing every change to the base model.

## Inspect the evidence
- [Evidence JSON](https://healthbenchprofessional.ai/evidence.json): Task definitions, dataset facts, scoring rules, source-version results, our interpretations, and reference IDs.
- [Sources](https://healthbenchprofessional.ai/sources/): Original papers and repositories with evidence locators.
- [Editorial method](https://healthbenchprofessional.ai/methodology/): Source reconciliation and interpretation boundaries.
- [About](https://healthbenchprofessional.ai/about/): Ownership and corrections.

Analysis updated: 2026-09-28