Updated
Updated · Computerworld · Sep 7
OpenAI Expands GPT-6 Astra After Messy Rollout Locked Out Paying Users for Days
Updated
Updated · Computerworld · Sep 7

OpenAI Expands GPT-6 Astra After Messy Rollout Locked Out Paying Users for Days

3 articles · Updated · Computerworld · Sep 7

Summary

  • Sam Altman apologized for a “messy” GPT-6 Astra launch after paying ChatGPT users could not access the model shortly after its September 4 debut.
  • Initial access was limited to organizations in OpenAI’s Daybreak cybersecurity program, leaving Plus, Pro, Business, Enterprise and many API users without the model despite launch messaging.
  • By September 5, OpenAI said Astra had reached Pro, Enterprise and Business Premium users in ChatGPT Work and Codex, and API access had opened; Plus and Business access was still rolling out over several days.
  • Analysts said the episode underscored that announced, available, entitled and production-ready are separate stages, warning enterprises to verify actual access rather than assume uniform rollout.
  • Gartner said Astra’s stronger reasoning, coding and automation could raise governance, cybersecurity and cost-control demands as companies test more autonomous workflows.

Insights

Did recent security breaches secretly force OpenAI to throttle Astra's release despite promises of immediate broad access?
What specific capabilities triggered Astra’s Critical security rating that forced OpenAI to lock out paying users?
How will the growing gap between AI marketing hype and actual enterprise readiness reshape future frontier model launches?