The GPT-5.6 model family, released on July 9, 2026, consists of three distinct tiers: **Sol**, **Terra**, and **Luna**. Unlike previous model versions that were often just different interfaces for the same underlying model, these are three separate models with different capability profiles, pricing, and safety configurations. ### The Three Tiers * **Sol (Flagship):** Designed for the most complex and demanding tasks. It is the most capable model in the family. * **Terra (Balanced):** Positioned as the balanced, mid-tier option, offering a compromise between capability and cost. * **Luna (Volume/Cost-Effective):** Optimized for speed, high-volume processing, and tasks with clear criteria. It is the most cost-effective model in the family. ### Key Differences * **Capability & Use Case:** Sol is intended for complex, ambiguous, or high-stakes reasoning. Luna is better suited for straightforward, large-volume tasks where efficiency is prioritized over deep reasoning. * **Safety & Safeguards:** The models feature different safety stacks calibrated to their specific capability profiles. For example, Sol and Terra include new activation classifiers for sensitive domains (such as biological or cybersecurity-related requests) that may cause the models to refuse certain high-risk prompts or require additional checks. * **Pricing:** The models are priced differently to reflect their tiers, with Sol being the most expensive and Luna the most affordable. * **Technical Implementation:** These models represent a shift toward a tiered strategy where the safeguard stack and reasoning modes (such as "Ultra" and "Max" modes) are integrated differently across the family to match the intended use case of each model. In summary, the choice between Sol, Terra, and Luna depends on whether you need maximum reasoning power (Sol), a balance of performance and cost (Terra), or high-speed, cost-effective processing for simpler tasks (Luna).