MazeBench: GPT-6.2 Sol narrows gap with Astra but still scores less at same price

patience_cave · x · 2026-11-20

Blogger patiencecave reports on his self-built MazeBench 3D open-world benchmark that GPT-6.2 Sol made strong progress over its predecessor and is approaching GPT-6.2 Astra's capabilities, though it still scores slightly worse than Astra at the same price. Accompanying maps show gpt-6.1-sol vs gpt-6.2-sol exploration traces, with 6.2 covering far more area. Self-built benchmark, unverified.

Original post →

More from Models

Models channel →