Models & research OpenAI August 10, 2026 · 10:00 UTC

Expanding Daybreak as the Cyber Defense Window Narrows

Meet GPT-5.6-Cyber, OpenAI’s cybersecurity-specific model available through Daybreak Red for authorized vulnerability research, exploit validation, and security testing.

Excerpt supplied by the publisher’s feed

Read the original article at OpenAI

Models & research Hugging Face Oct 7, 2026

Multimodal open d1 decision models for the edge

Read brief Original at Hugging Face ↗

Models & research GitHub Copilot Changelog Oct 7, 2026

Purpose-built model for leaked secret detection

Secret protection should keep pace with the way you build software, whether you write code yourself or work with an AI agent. With our new purpose-built model, we’re bringing context-aware… The post Purpose-built model for leaked secret detection appeared first on The GitHub Blog.

Read brief Original at GitHub Copilot Changelog ↗