OpenAI has launched GPT-5.6, its latest flagship model, in a restricted rollout coordinated with the U.S. government. The release marks a significant shift in how the company structures its AI offerings, introducing a three-tier naming system that separates capability levels from generation numbers. Access currently remains limited to trusted partners, but OpenAI plans to open the models to general availability within weeks.

What Are the Three GPT-5.6 Tiers, and How Do They Differ?

Rather than releasing a single monolithic model, OpenAI is rolling out GPT-5.6 as three distinct versions, each optimized for different use cases and performance requirements. The company named them after celestial bodies to make the tier system more intuitive for users and developers.

• Sol: The flagship model designed for maximum capability and reasoning power, built for long-horizon agentic work, stronger reasoning, and cybersecurity applications.
• Terra: A balanced version that sits between Sol and Luna, optimized for everyday work and general-purpose tasks without sacrificing too much speed or cost.
• Luna: The fastest and most affordable tier, prioritized for speed and cost efficiency while maintaining reasonable capability for lighter workloads.

This dual-axis naming system represents a departure from OpenAI's previous approach. The number (5.6) indicates the model generation, while Sol, Terra, and Luna mark capability tiers that advance independently. This structure gives users and developers clearer choices across intelligence, speed, and cost without forcing them to choose between a single option or navigate confusing variant names.

Why Is the Rollout Restricted, and When Will It Be Widely Available?

OpenAI describes the current phase as a "limited preview" tied directly to coordination with the U.S. government. The company briefed government officials on the models' capabilities before launch and, at the government's request, first opened access to a small group of trusted partners whose participation it disclosed to authorities.

This cautious approach reflects growing regulatory scrutiny around advanced AI systems. Rather than a traditional public beta, OpenAI is treating this rollout as a collaborative process with government oversight. The company plans to continue testing and coordinating with partners throughout the preview period before moving to broader access. According to OpenAI's timeline, Sol, Terra, and Luna should reach general availability within weeks of the June 26, 2026 launch date.

How Does GPT-5.6 Compare to Earlier Versions?

GPT-5.6 represents the latest step in a rapid cadence of model releases. Just two months earlier, in April 2026, OpenAI launched GPT-5.5, which introduced agentic coding, deep research capabilities, and computer use features that allow the model to operate software interfaces without step-by-step instructions.

The progression from GPT-5.5 to GPT-5.6 builds on these foundations. GPT-5.6 is specifically engineered for long-horizon agentic work, meaning it can handle extended, multi-step tasks that require sustained reasoning and planning. It also brings stronger reasoning capabilities and enhanced cybersecurity features, suggesting OpenAI is targeting enterprise and security-focused use cases alongside general developer needs.

Steps to Choose the Right GPT-5.6 Tier for Your Use Case

Selecting between Sol, Terra, and Luna depends on your specific requirements. Here's how to match each tier to common scenarios:

• Choose Sol if: You need maximum reasoning power, are building long-running autonomous systems, or require advanced cybersecurity analysis. Sol is the best fit for complex, multi-step problems that demand the highest capability.
• Choose Terra if: You're doing everyday work, building standard applications, or need a good balance between capability and cost. Terra is ideal for most developers and businesses that don't need the absolute top-tier performance.
• Choose Luna if: Speed and cost are your primary concerns, you're handling lighter workloads, or you're building high-volume applications where latency matters. Luna is the most economical choice for straightforward tasks.

OpenAI's broader version history shows a consistent pattern of expanding capability across reasoning, coding, and multimodal understanding. Since ChatGPT's launch in November 2022, each major release has pushed forward on these fronts. GPT-5.5, released in April 2026, added unified handling of text, images, audio, and video through a single system, replacing the earlier approach of stitching together separate systems for each format. GPT-5.6 builds on this foundation with enhanced reasoning and agentic capabilities.

The three-tier structure also reflects a broader industry trend toward offering multiple capability levels rather than a single flagship model. This approach gives developers and enterprises more flexibility to optimize for their specific constraints, whether those are computational budgets, latency requirements, or capability needs. As OpenAI moves from limited preview to general availability, the Sol, Terra, and Luna naming system will likely become familiar to millions of ChatGPT users and API developers.