GPT-6 Astra reportedly scores 100% on ExploitBench, finds two zero-days in testing

VraserX · x · 2026-09-04

According to the post, GPT-6 Astra scored 100% on ExploitBench and discovered two previously unknown zero-day vulnerabilities during evaluation. The public release will reportedly be deliberately blocked from some advanced cyber tasks. The author frames this as the start of a new AI era — capabilities that exist but most people are simply not allowed to access. (Unverified claim.)

Original post →

More from Models

Models channel →