GPT-6.2 Astra crushes Opus 5.5 on math benchmarks at a fraction of the cost

VraserX · x · 2026-11-18

On two math-proof benchmarks, GPT-6.2 Astra far outpaces Anthropic's Claude Opus 5.5 and Fable 5.2:

The poster notes Astra costs a fraction of either and calls it not even close.

Related event: GPT-6.2 Astra retakes MathArena top spot with big lead(2 posts)→

Original post →

More from Models

Models channel →