HealthBench
Rubric-graded benchmark of realistic health conversations written with physicians, scoring the quality and safety of model responses.
Actively evaluated by Artificial Analysis on current frontier models; no saturation evidence recorded.
Performance Timeline
Longitudinal progression of model scores against human baselines.Performance & Historical Trajectory
Empirical score progression across model release dates and evaluation rounds.
| Model | Score | Date | Source Type | Provenance |
|---|---|---|---|---|
| Claude Sonnet 5.5 | 62.1% | 2026-11-15 | independent | Source ↗ |
| GPT-6.1 Sol | 58.5% | 2026-11-15 | independent | Source ↗ |
| Claude Fable 5.2 | 59.6% | 2026-11-15 | independent | Source ↗ |
| GPT-6 Luna | 54.6% | 2026-11-15 | independent | Source ↗ |
| GPT-6 Astra | 58.2% | 2026-11-15 | independent | Source ↗ |
| Claude Haiku 5.5 | 58.5% | 2026-11-15 | independent | Source ↗ |
| Grok 4.7 | 56% | 2026-11-15 | independent | Source ↗ |
| Claude Opus 5.5 | 57% | 2026-11-15 | independent | Source ↗ |
| Gemini 4 Argon | 49.1% | 2026-11-15 | independent | Source ↗ |
| Gemini 3.8 Flash | 48.4% | 2026-11-15 | independent | Source ↗ |
| GLM 5.3 | 54.5% | 2026-11-15 | independent | Source ↗ |
Human Baseline & Difficulty Horizon
Calibrated human reference points, specialist benchmarks, and ceiling thresholds.No peer-reviewed human baseline has been published for this benchmark.
Metric & Scoring Methodology
Verification protocols, aggregation formulas, and specialized metric variants.rubric score (%)Dataset & Compute Cost
Evaluation volume, public availability, API pricing, and local hardware requirements.$5 – $20 USD for full benchmark evaluation run on frontier APIs.
1x NVIDIA RTX 4090 (24GB) or A100 (40GB/80GB) via vLLM / SGLang
How to Run & Reproduce
Standardized evaluation protocols, CLI commands, and reproducible runner templates.lm_eval --model hf --model_args pretrained=<model_path> --tasks healthbench --batch_size autoopencompass --datasets healthbench --models <model_config># Standard API Evaluation Loop
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": prompt}],
temperature=0.0,
)When publishing results for HealthBench, always report the exact prompt template, few-shot exemplar ordering, sampling temperature (temperature=0), maximum reasoning budget tokens, and the precise timestamped model snapshot ID.
Contamination & Memorization Analysis
Audit of pretraining exposure risks, memorization vectors, and refresh policies.Static fixed snapshot
Public on web / HuggingFace