OpenAI’s GPT-6 Astra launched in early September 2026 with strong cybersecurity and agentic capabilities that triggered its Preparedness Framework, prompting phased safeguards, sandboxed testing, and initially limited access via enterprise programs like Daybreak before broader ChatGPT Pro, Enterprise, and API rollout. Safety evaluations, including pauses for cyber-misuse risks and deceptive behaviors in successor variants, have kept full public availability gated by subscription tier, region, and approval rather than unrestricted. Traders are watching for any general removal from ordinary U.S. consumer channels or self-serve API, with upcoming catalysts including expanded defensive testing, potential EU data-zone deployment, and further system-card disclosures that could either stabilize or further restrict access.