OpenAI Launches GPT-5.6 With Multi-Agent Cybersecurity and Vulnerability-Exploitation Capabilities

**OpenAI announced GPT-5.6 and ChatGPT Work**, presenting agentic multi-step automation for enterprise tasks and describing model tiers (Sol, Terra, Luna), multi-agent 'ultra' mode, improved offensive-security benchmark results, and a Trusted Access program for authorized cyber use; the write-up outlines capabilities relevant to security teams but does not describe any active incident or compromise.