An AI research company has initiated a limited preview of its next-generation GPT-5.6 model series, featuring three distinct tiers: Sol, Terra, and Luna. This phased release begins with a select group of trusted partners, a step taken in coordination with the U.S. government. The company plans a broader public rollout in the coming weeks, signaling a significant advancement in its AI offerings.
New Models and Pricing Structure
The GPT-5.6 series introduces a new naming system, with each model catering to different needs and budgets. Sol is positioned as the flagship model for complex tasks, while Terra offers a balance of performance and cost. Luna is designed as the most affordable and fastest option, providing strong capabilities at the company's lowest price point.
This tiered structure provides clearer choices for developers, with Terra delivering performance competitive with the previous GPT-5.5 model at half the price. Pricing is set per million tokens, with Sol at $5 input and $30 output, Terra at $2.50 and $15, and Luna at $1 and $6. This strategic pricing aims to broaden access to powerful AI tools for a wider range of applications.
Advancements in Reasoning and Performance
The flagship model, GPT-5.6 Sol, is described as the company's most capable to date, with significant improvements in agentic capabilities. It introduces a "max reasoning effort" feature for deeper analysis and a new "ultra mode" that leverages subagents. The models also demonstrate state-of-the-art performance across benchmarks in coding, biology, and cybersecurity, showing broad improvements over previous versions.
A Focus on Cybersecurity and Safeguards
A central focus of the GPT-5.6 launch is a significantly enhanced safety and security framework. The company has implemented its most robust safeguards to date, designed to hold up against adversarial pressure while preserving legitimate uses. This layered approach includes model-level protections, real-time misuse classifiers, and account-level monitoring to prevent misuse.
The models were developed to make prohibited offensive cyber activity more difficult and detectable without limiting beneficial defensive work. In internal evaluations, GPT-5.6 Sol did not cross the company's "Cyber Critical" threshold, and extensive automated and human red-teaming was conducted. This demonstrates a commitment to balancing capability with responsible deployment and mitigating potential risks.
Phased and Coordinated Release
The initial release is a limited preview for trusted partners, a decision made in consultation with the U.S. government. The company described this as a short-term measure while it works with the administration to develop a framework for future model releases. Broader availability for ChatGPT, Codex, and API users is expected to follow in the upcoming weeks.
The introduction of the GPT-5.6 series represents a strategic evolution, combining advanced capabilities with a tiered, cost-effective structure. The company's emphasis on a robust, multi-layered safety system and a cautious, government-coordinated rollout underscores the importance of responsible AI deployment. This preview sets the stage for a significant expansion of accessible and powerful AI tools for developers and enterprises.