Mystery model "Astra" reportedly beats 5.6 Sol Pro on FrontierMath T4

ctjlewis · x · 2026-09-04

X user FateOfMuffins flags to AI safety researcher Ryan Greenblatt that a mysterious model labeled "Astra (NONE)" scores higher than "5.6 Sol Pro (Max)" on the FrontierMath T4 benchmark. The reposter reacts with "they fucking cooked." The model's identity is unknown and the claim is unverified.

Original post →

More from Models

Models channel →