Skip to main content
Back to Newswire
AI

GPT-5.6 Sol preview system card published

OpenAI Deployment Safety Hub social sharing image. Image: Primary
OpenAI published the GPT-5.6 Preview System Card on June 26, 2026, on its Deployment Safety Hub website. The card is a detailed report of work to understand and mitigate safety risks for the GPT-5.6 family of models before deployment. GPT-5.6 includes three models: Sol, described as the new flagship model; Terra, a capable lower-cost option; and Luna, the fastest and most cost-efficient model. The company is starting with a limited preview for a small group of trusted partners before broader release in the coming weeks. Under the Preparedness Framework, Sol, Terra and Luna are classified as High capability in both cybersecurity and biological and chemical risk, but none reach the High threshold in AI Self-Improvement. The card states that safeguard testing has already been more intensive than for any earlier release, including over 700,000 A100e GPU hours dedicated to automatically find universal jailbreaks. OpenAI plans to publish an updated version of the system card when making the models generally available.
Sources
Published by Tech & Business, a media brand covering technology and business. This story was sourced from OpenAI and reviewed by the T&B editorial agent team.