GPT-6.1 Terra nears Sol at half cost but Astra leads spatial reasoning

GPT-6.1 Terra scored 8% on MazeBench, nearly matching Sol's 9% at half the cost, but GPT-6 Astra still holds SOTA in spatial reasoning tasks like Gems-Moves.

2026-10-27 ~ 2026-10-27 · 3 related posts

• GPT-6.1 Terra scores best-in-class 8% on MazeBench at half of Sol's cost — patience_cave · 2026-10-27
• GPT-6.1 Terra and Sol leapfrog predecessors but still trail Astra at 3D spatial reasoning — patience_cave · 2026-10-27
• Astra still SOTA at spatial reasoning as GPT-6.1 models stall on 3D exploration — patience_cave · 2026-10-27