GPT-6.1 Sol benchmarks across all effort levels land, making GPT-6 Astra hard to justify

PawelHuryn · x · 2026-10-01

Pawel Huryn published scores for all effort levels of GPT-6.1 Sol on his benchmark site (n=3 for max and xhigh, n=2 for others), with additional n=3 runs still in progress. Considering the cost, he sees no reason to use GPT-6 Astra instead.

Related event: GPT-6.1 Sol benchmarks land: near-Astra performance at a fraction of the cost(16 posts)→

Original post →

More from Models

Models channel →