OpenAI reportedly launches GPT-6 'Astra' with 'critical' cyber rating
GooseberryGOLD · reddit · 2026-09-05
According to a Verity News report (not yet officially confirmed by OpenAI), the company has launched its next-gen model GPT-6 "Astra," which reportedly received a "critical" rating on cyber capability safety evaluations — the highest tier of security concern. Details and methodology remain unverified.
Related event: OpenAI launches GPT-6 Astra with first-ever Critical cybersecurity rating(3 posts)→
More from Models
- Zvi: If Models Can Do This, Steganographic Output Only Needs a Convention — TheZvi · 2026-09-05
- RedMonk: Open Weight Models Already Match the Capability That Changed the Industry — rseroter · 2026-09-05
- Benchmarked 21 Qwen3.8-27B quants on 16GB VRAM: bartowski IQ4_XS wins — Storterald · 2026-09-05
- First hands-on: GPT-6 Astra nails Blender modeling of a prison phone in one pass — AIandDesign · 2026-09-05
- GPT-6 Astra appears in Codex during testing, availability scope unclear — dotey · 2026-09-05
- GPT-6 Astra spotted in early tests, reportedly beating Fable benchmark — Angaisb_ · 2026-09-05