OpenAI Agents API adds Daybreak cyber access, docs reveal gpt-6-sol and gpt-5.6-cyber models

nikunjhanda · x · 2026-11-21

OpenAI's Agents API now supports Daybreak access (Daybreak Blue and Daybreak Red) for building long-running security agents that continuously monitor risks. Per documentation screenshots, sessions specify accessprograms.cyber — with model gpt-6-sol for Daybreak Blue and gpt-5.6-cyber for Daybreak Red, two previously unannounced model codenames. Use cases include investigating suspicious activity, implementing security updates, verifying fixes, triaging for human review, and quarantining suspicious actors. Access requires applying to the Daybreak program.

Related event: OpenAI Expands Daybreak Security Model Access and Agents API Integration(2 posts)→

Original post →

More from Models

Models channel →