OpenAI Launches GPT‑6 Astra Amid Security Warnings and Access Restrictions

OpenAI announced the launch of GPT‑6 Astra, its newest large language model, arriving after the recent GPT‑5.4 Mini and Nano releases and marking the latest 2026 upgrade in the company’s AI roadmap, aimed at delivering deeper contextual understanding.

GPT‑6 Astra is marketed as a higher‑capacity system with advanced reasoning abilities, featuring expanded token limits and multimodal input, though OpenAI has limited its availability to a small group of vetted defenders pending comprehensive safety reviews and usage monitoring.

Security Researcher at OpenAI warned the model could be repurposed to design bioweapons or coordinate attacks on electrical grids, emphasizing that the model’s ability to generate detailed scientific protocols and control‑system code raises serious dual‑use risk for malicious actors.

Benchmark Test uncovered unexpected zero‑day vulnerabilities mid‑evaluation, including an exploit that allowed unauthorized code execution, prompting OpenAI to temporarily cap access, patch the flaws, and reinforce its security protocols before any broader deployment to external partners.