OpenAI has officially introduced its next-generation artificial intelligence family, the GPT-5.6 series. This new lineup features three distinct variants—Sol, Terra, and Luna—each designed to cater to different levels of computing power, speed, and cost. While these models represent a significant leap in reasoning, coding, and cybersecurity capabilities, their release follows a unique, staggered rollout strategy coordinated closely with the U.S. government.
At the top of the hierarchy sits Sol, the new flagship model. OpenAI positions Sol as its most capable and powerful tool to date, specifically engineered to tackle the most complex challenges in software engineering, scientific research, and advanced cybersecurity. The model introduces enhanced reasoning features, including a new “max” effort setting that allows it to process tasks more deeply, and an “ultra” mode that leverages sub-agents to break down massive problems into manageable steps.
Complementing the flagship is Terra, which OpenAI describes as a balanced model built for everyday business tasks. Terra is designed to provide high-level performance that remains competitive with the previous GPT-5.5 series, yet it does so at roughly 50% of the cost. This makes it an attractive middle-ground option for enterprises looking to scale their AI operations without a steep increase in expenses.
The third variant, Luna, rounds out the family as the fastest and most cost-efficient member of the series. While Luna is optimized for speed and high-volume work like drafting and routine automation, it still retains strong functional capabilities. Early insights suggest that while Luna is the most affordable in the lineup, it performs remarkably well in tasks requiring rapid response times and efficient token usage.
This launch is not a standard open-access rollout. Instead, OpenAI is currently limiting access to these models to a small group of approximately 20 trusted partners and organizations. This cautious approach comes at the request of the U.S. government, which is currently exercising more oversight over high-capability AI models. Under an executive order signed earlier this month, federal agencies are reviewing the potential risks—particularly regarding cybersecurity and biological research—before these powerful tools reach the general public.
OpenAI has explicitly stated that it does not view this government-managed access process as a long-term solution. In its official communications, the company noted that such barriers can prevent developers, researchers, and cyber defenders from accessing the tools they need to strengthen digital infrastructure. Despite these constraints, OpenAI remains optimistic, with plans to make Sol, Terra, and Luna generally available to the public and developers in the coming weeks.
Safety has been a primary focus throughout the development of the GPT-5.6 series. OpenAI spent weeks pressure-testing the system, specifically hardening it against real-world adversarial attacks and unauthorized cyber activities. According to the company’s internal reports, while the models are highly proficient at identifying software vulnerabilities, they do not currently possess the ability to perform autonomous, end-to-end attacks against hardened targets.
The industry is closely watching how these models perform in real-world scenarios. Early benchmarks indicate that the flagship Sol variant performs exceptionally well on complex tasks like Terminal Bench 2.1. By offering three clear tiers—Sol for heavy-duty research, Terra for balanced work, and Luna for speed—OpenAI is attempting to streamline the way users choose the right tool for their specific needs, moving away from confusing branding toward a more logical, generational naming system.
As the tech world awaits the broader release, the current limited preview serves as a testing ground for both the technology and the new regulatory framework. For now, the integration of these models into professional workflows remains restricted, but the rapid pace of development suggests that the next few weeks will bring significant changes to how businesses and developers leverage AI for their most demanding projects.
