Skip to content
K = 0.731

Artificial Analysis Launches HealthBench and HealthBench Professional Leaderboards

1 reports 1 reporting sources 2 hours agoNew

Understand the event

AI overview

Artificial Analysis launched leaderboards for OpenAI’s HealthBench and HealthBench Professional, with GPT-6 Astra (max) leading the length-adjusted Professional ranking at 66.0% and Claude Sonnet 5.5 (max with fallback) leading HealthBench at 62.1%. Both benchmarks evaluate AI responses to realistic health conversations using physician-created rubrics, and the leaderboards report raw and length-adjusted scores. The length adjustment penalizes responses exceeding 2,000 characters and boosts shorter responses, which can substantially change the rankings. Claude Opus 5.5 (max with fallback) has the highest raw HealthBench Professional score at 71.4%, but falls to sixth place at 59.9% after adjustment.

Generated by AI from reports · Updated 1 hours ago

Latest development10/20 07:03 (UTC+8)
Artificial Analysis launches HealthBench and HealthBench Professional leaderboardsArtificial Analysis · X

Models in this story

Compare their API pricesAPI prices of all models

Report timeline

Follow the reports to explore different sides of the event.

10/20
  1. Artificial Analysis · X
    Artificial Analysis launches HealthBench and HealthBench Professional leaderboards

    Artificial Analysis is launching leaderboards for OpenAI’s HealthBench and HealthBench Professional, reporting raw and length-adjusted scores on physician-created rubrics for realistic health conversations. GPT-6 Astra (max) leads HealthBench Professional at 66.0%, while Claude Sonnet 5.5 (max with fallback) leads HealthBench at 62.1%. Length adjustment penalizes replies exceeding 2,000 characters and boosts shorter replies; Claude Opus 5.5 (max with fallback) has the highest Professional raw score at 71.4% but drops to sixth at 59.9% after adjustment.

Tomorrow at 08:00, a briefing like this arrives in your inbox.

The same kind of briefing as the one above: a few stories worth reading. Free. Mail starts after you confirm, and every letter has a one-click unsubscribe. Read a past issue

Event heat trend

There is not enough continuous observation data to plot a trend yet.

Keep reading

Latest daily · Issue 27TorchNeuron brings native PyTorch support for Hugging Face models on AWS Trainium