GPT-5.6 Breaks Records on Coding Leaderboards

OpenAIDevs · x · 2026-07-10

OpenAI highlights the superior token efficiency of GPT-5.6. On the Artificial Analysis Coding Agent Index, Sol achieved a score of 80 using the highest reasoning setting, utilizing 54% fewer output tokens and completing tasks 57% faster than the runner-up model. In knowledge work evaluations, Terra outperformed GPT-5.5 at a lower cost, while Luna approached the peak performance of GPT-5.5 at less than half the estimated API cost.

Related event: OpenAI Launches GPT-5.6 Series: Multi-Agent and Cost-Efficiency(119 posts)→

Original post →

More from coding & agent

coding & agent channel →