GPT-6 Luna beats GPT-5.6 Sol on HealthBench Professional at ~34x lower cost

BorisMPower · x · 2026-09-24

OpenAI's Karan Singhal introduced GPT-6 Luna and Sol as a step toward health intelligence "too cheap to meter." On HealthBench Professional—built on real clinician tasks like care consultations and medical documentation—GPT-6 Luna outscores GPT-5.6 Sol, their best model weeks ago, at 34x lower cost: roughly a third of a cent per response.

Original post →

More from Models

Models channel →