Leaked eval: GPT-6 Sol scores 1.6x GPT-5.6 Sol on robotics, 47% cheaper but off Pareto frontier

ycombinator · x · 2026-09-25

An unverified eval leak claims GPT-6 Sol scores 1.6x as high as GPT-5.6 Sol on robotics tasks while being 47% cheaper, though it still sits below the Pareto frontier under Opus 5.5. Authenticity remains unconfirmed.

Original post →

More from Models

Models channel →