OpenAI's GPT-5.6 Outperforms Physicians in Healthcare Tasks, Reduces Cost by 25x

The new GPT-5.6 Luna variant outperforms GPT-5.5 at 25 times less cost, while the largest variant, GPT-5.6 Sol, sets a new high bar for cost-performance. Physicians found fewer flaws in GPT-5.6 responses than in physician-written responses across 20,000 axis ratings in diverse patient and clinician-facing healthcare tasks.