WHAT NEW IN X11.07 SAT

AI collections from X · 25 collections · 72,532 posts

gpt-6POST

GPT-6 Astra High scores 77.34 on PPTBench

PPTBench reports Astra High at 77.34, but about one in five tasks still fails a gate, showing visual coding agents still have clear gaps.

MR. SHAHBAZ
@Mr_Shahbaz_AiArchived 2026.09.30Originally posted 2026.09.29
MR. SHAHBAZ @Mr_Shahbaz_AiOriginally posted 2026.09.29
Open original on X

GPT-6 Astra High leads the reported PPTBench score at 77.34, but around one in five tasks still fails a gate. PPTBench breaks those failures down clearly, starting with semantics before checking render/text quality and details. That level of evaluation gives a much clearer picture of what visual coding agents still need to improve.

This post quotes another post: x.com/einsiaai/status/2104941770433814889

x.com/mr_shahbaz_ai/status/2104953317747282111Archived 2026.09.30 10:24