OpenAI today launched a limited preview of its GPT-5.6 series, which includes the flagship Sol, a balanced daily work model named Terra, and Luna, a fast and affordable model.
Terra has similar performance to GPT-5.5 but is 2x cheaper, and Luna offers “high capacity” at OpenAI’s lowest price. GPT-5.6 Sol is the most powerful OpenAI model to date, with agent enhancements in coding, biology, and cybersecurity. There is a new “max” reasoning effort and an “ultra” mode that uses subagents for complex work.
OpenAI claims that GPT-5.6 Sol has its “most robust security stack to date” with protections for high-risk activities, sensitive cyber requests, and misuse. It has been tested for weaknesses and hardened against real-world attacks. OpenAI says the safeguards allow the model to provide “substantial benefits for legitimate defensive work” while limiting prohibited offensive uses. According to OpenAI, it is better to help users find and fix vulnerabilities rather than carrying out end-to-end attacks.
The Trump administration is restricting the launch of GPT-5.6 and OpenAI has agreed not to release it to all users. The model is currently available to a small group of trusted partners, but OpenAI plans a wider launch after more extensive testing. In its GPT-5.6 announcement, OpenAI pushed back on the administration’s request to withhold the model and said the government’s process for accessing AI should not become the default in the long term.
We do not believe that this type of government access process should become the default solution in the long term. It curates the best tools for users, developers, businesses, cyber defenders, and global partners who need them. We are taking this action in the near term because we believe it is the strongest path toward broader availability in the coming weeks as we work with the Administration to develop the Cyber Executive Order framework and a repeatable process for future model releases.
The Trump administration is establishing a process for benchmarking and evaluating new AI models before their launch, according to a June 2 executive order. The administration previously forced Anthropic to remove access to Claude Fable 5 and Mythos 5, even though Anthropic adhered to a voluntary government review process and added guardrails based on government feedback.
GPT-5.6 models are available via the API and Codex to a range of trusted OpenAI partners and organizations. OpenAI says they will “soon” be available more widely in ChatGPT, Codex, and the API.
Note: Due to the political or social nature of the discussion on this topic, the discussion thread is located in our political news forum. All forum members and site visitors are welcome to read and follow the thread, but posting is limited to forum members with at least 100 posts.
