OpenAI has begun a limited preview of its GPT-5.6 model family, led by the flagship Sol model and joined by the lower-cost Terra and Luna variants. The company says the early release is restricted to trusted partners while it continues safety testing ahead of broader availability.
OpenAI has started a limited preview of GPT-5.6, a new family of models comprising Sol, Terra and Luna. In its June 26 announcement, the company described Sol as its flagship model, Terra as a balanced option for everyday work, and Luna as a faster, more affordable model.
OpenAI said Terra offers performance competitive with GPT-5.5 at half the cost, while Luna is intended to provide capable access at the family’s lowest price point. The announcement did not include detailed benchmarks, methodologies or comparative task results supporting those performance and cost claims.
OpenAI’s API documentation describes GPT-5.6 Sol as the family’s frontier model for complex professional work. The documentation says Sol is available through the API and lists supported tools and pricing information, though developers’ practical access may depend on the terms of the preview.
OpenAI said GPT-5.6 Sol is launching with its “most robust safety stack to date.” According to the company, it has strengthened safeguards for higher-risk activity, sensitive cyber requests and repeated misuse.
The company also said it spent multiple weeks identifying weaknesses, pressure-testing its systems and hardening the models against real-world attacks before opening the preview. OpenAI’s GPT-5.6 Preview System Card covers Sol, Terra and Luna, including capability assessments, pre-deployment mitigations and the limited-preview deployment plan.
The available announcement provides only a high-level account of the testing. It does not identify all of the evaluation outcomes, disclose a complete set of benchmark measurements, or provide independent evidence of how the stated mitigations perform in use. As a result, external assessment of the models’ capabilities and safeguards remains limited to the information OpenAI has published.
OpenAI said it intends to make Sol, Terra and Luna generally available in the coming weeks. Before the announcement, the company said it had discussed the planned release and the models’ capabilities with the U.S. government as part of ongoing engagement.
At the government’s request, OpenAI said the initial preview is limited to a small group of trusted partners whose participation has been shared with the government. OpenAI said it will continue testing the models and coordinating with those partners before broadening access.
The company argued that this kind of government-access arrangement should not become a long-term default for releases. It said extended restrictions could delay access for developers, enterprises, cyber defenders and global partners that need advanced AI tools.
For now, GPT-5.6 is being presented as a staged deployment rather than a full public launch. OpenAI has set an expectation of wider availability in the coming weeks, but it has not published a specific general-release date or detailed eligibility criteria for access during the preview.
OpenAI introduces the GPT 5.6 family in a limited release OpenAI has started a limited preview of GPT 5.6, a new family of models comprising Sol , Terra and Luna .
In its June 26 announcement, the company described Sol as its flagship model, Terra as a balanced option for everyday work, and Luna as a faster, more affordable model.
OpenAI said Terra offers performance competitive with GPT 5.5 at half the cost, while Luna is intended to provide capable access at the family’s lowest price point.
Continue reading
