GPT-6 Sol Gains 2 Points in Coding Agent Index at Half the Cost

ArtificialAnlys · x · 2026-09-23

Follow-up from Artificial Analysis: in OpenAI's Codex harness, GPT-6 Sol (max) scores 57 in the Coding Agent Index (+2 over GPT-5.6) at roughly half the cost per task, while Luna (max) regresses 2 points. A sub-finding of its full GPT-6 evaluation.

Related event: GPT-6 Sol divides reviewers: gains in coding agents, regression on DeepSWE, at half the price(11 posts)→

Original post →

More from Models

Models channel →