OpenAI releases GPT-5.6 Sol, Terra and Luna as a Limited Preview
OpenAI today released a limited preview of its GPT-5.6 series, which includes the flagship model Sol, a balanced everyday model called Terra, and Luna, a fast and affordable model.
Terra offers comparable performance to GPT-5.5 but at half the price, while Luna delivers “strong performance” at OpenAI’s lowest price. GPT-5.6 Sol is OpenAI’s most powerful model to date, featuring agent-based enhancements in coding, biology, and cybersecurity. It includes a new “Max” mode for maximum computational effort and an “Ultra” mode that utilizes subagents for complex tasks.
According to OpenAI, GPT-5.6 Sol features the most robust security mechanism to date, with safeguards against high-risk activities, sensitive cyber requests, and abuse. It has been vulnerability tested and hardened against real-world attacks. OpenAI states that these safeguards allow the model to provide significant value for legitimate defensive tasks while limiting prohibited offensive use. It is better suited for detecting and remediating vulnerabilities than for conducting end-to-end attacks, according to OpenAI.
Trump administration restricts GPT 5.6 access
The Trump administration is restricting the release of GPT-5.6, and OpenAI has agreed to hold off on making it available to all users for the time being. The model is currently only available to a small group of trusted partners. However, OpenAI plans a wider rollout after further testing. In its announcement regarding GPT-5.6, OpenAI rejected the administration’s request for restraint, stating that government access to AI should not become the long-term standard.
“We believe this government access procedure should not become the permanent standard. It keeps the best tools out of reach of users, developers, businesses, cybersecurity experts, and global partners who desperately need them. We are taking this short-term measure because we believe it is the most effective way to achieve broader availability in the coming weeks, while we work with the government to develop the framework for the cybersecurity regulation and a repeatable process for future model versions.”
The Trump administration, pursuant to an order issued on June 2, is developing a process for evaluating and comparing new AI models before their market launch. This follows the administration’s decision to force Anthropic to block access to Claude Fable 5 and Mythos 5, despite Anthropic’s adherence to a voluntary government review process and its implementation of additional security measures based on government feedback.
The GPT 5.6 models are available via the API and Codex to a select group of trusted OpenAI partners and organizations. According to OpenAI, broader access via ChatGPT, Codex, and the API will follow shortly.
