0

MedScribe

Frontier

Can models support doctors with their administrative work?

Publisher
Vals AI
Domain
Medical
Published
Jun 2026
Official leaderboard
vals.ai/benchmarks/medscribe

Cite

Notes

Only stored in your browser.

Attribution

Leaderboard scores
Vals AI
Attribution policy →

Top score 88.1% by GPT-5.1 - 47 models reporting (23 frontier)

Score history

44
50%63%75%88%100%Apr 25Jul 25Oct 25Jan 26Apr 26o3Gemini 2.5 FlashGPT-5Claude Sonnet 4.5GPT-5.1

Top models

47
MedScribeBar chart with 21 bars. Highest value: GPT-5.1 at 88.1.
21 models

Where it's ranked

1

FAQ

What is MedScribe?
Can models support doctors with their administrative work?
What is the current top score on MedScribe?
The top reported score is 88.1% by GPT-5.1, across 47 models reporting (23 from frontier labs).