Safety overview: GPT-6 Astra

Must Read

OpenAI Blog · 2026-09-03 00:00 UTC AI

9.5
Relevance
9
Importance

AI Summary

GPT-6 Astra is described as the most capable broadly deployed model to date. It is the first model to achieve a Critical level of cybersecurity capability under the Preparedness Framework.

AI analysis by Gemini

Why should I care?

This represents a significant advancement in frontier foundation LLM capabilities and cybersecurity milestones.

Content

GPT-6 Astra is our most capable broadly deployed model and our first to reach the Critical level of cybersecurity capability under our Preparedness Framework.


Your feedback

Keep this article

Protects it from automatic cleanup.