gpt-6IMAGE
GPT-6.1 Sol scores 9% on MazeBench
On the MazeBench maze benchmark, GPT-6.1 Sol scored 9%, just above Opus 5.5, retaining many of Astra's puzzle-solving abilities at a tenth of the cost.
💺
@patience_caveArchived 2026.10.02Originally posted 2026.10.02
💺 @patience_caveOriginally posted 2026.10.02
MazeBench vs GPT-6.1 Sol GPT-6.1 Sol scored 9%, scoring just above Opus 5.5 Sol retained many of the puzzle solving capabilities from GPT-6 Astra but at 10% the cost
x.com/patience_cave/status/2105695074017526125Archived 2026.10.02 10:05