Observed arrival · 2026-09-25
The Examination Room puts AI through a clinical viva
A clinician-written benchmark tests whether AI models gather evidence, order tests, and defend treatment plans under time and budget constraints.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- AI evaluation researchers and clinical safety teams
- Worth noticing
- A private notebook stays hidden from three examiners while the model works through a ₹45,000 budget and 25-minute clock.
Field notes
The environment withholds case facts until a model requests them, then tracks tests against a ₹45,000 budget and a 25-minute clock. Three examiner agents can challenge a plan for up to 65 turns; an event ledger records actions, while two clinicians review runs blind, then compare marks and settle disagreements. The reported sample is 24 runs across six models, all using one clinician-written dental case.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue