OpenAI has announced the release of GPT-5.6, its new family of language models, just two months after GPT-5.5. However, the upgrade for ChatGPT and Codex will not be immediate for all users. The company has opted for a staggered rollout, initially granting access to only 20 trusted partners, following an explicit request from the United States government.
The three versions of GPT-5.6: Sol, Terra, and Luna
GPT-5.6 consists of three models inspired by celestial bodies: Sol, Terra, and Luna. OpenAI explains that the number identifies the generation, while the names indicate durable capability tiers, each with its own development cadence. Sol is the most powerful model, optimized for complex reasoning tasks, while Terra and Luna offer a balance between performance and cost. This modular structure allows developers and enterprises to choose the solution that best fits their needs, balancing intelligence, speed, and expense.
Sponsored Protocol
New reasoning options: max and ultra
With GPT-5.6 Sol, OpenAI introduces two new reasoning modes: "max" and "ultra." The max mode gives the model maximum time to produce deep answers, while the ultra mode goes beyond the capabilities of a single agent by leveraging subagents to accelerate complex work. According to the company, the model shows significant advances in coding, biology, and cybersecurity, fields where precision is crucial.
Partnership with the US government for controlled release
OpenAI stated that it collaborated with the US government before the launch, sharing the plans and model capabilities. At their request, the preview is limited to a small group of trusted partners, whose participation has been disclosed to the authorities. During this phase, OpenAI will continue testing and coordinating with partners to ensure safety and reliability. General access is expected in the coming weeks, but no specific date has been set.
Sponsored Protocol
Token pricing and predictable caching
GPT-5.6 pricing is per million tokens: Sol costs $5 input and $30 output; Terra $2.50 and $15; Luna $1 and $6. OpenAI also introduces more predictable prompt caching, with explicit cache breakpoints and a minimum cache life of 30 minutes. Cache writes are billed at 1.25 times the model's uncached input rate, while cache reads receive a 90% discount. This pricing structure aims to make costs more transparent for developers integrating the API.
Safety and safeguards in new models
OpenAI emphasizes that GPT-5.6 Sol, Terra, and Luna come with the most robust safeguards ever implemented, with configurations tailored to each model's capabilities. The company has heavily invested in preventing misuse, especially in sensitive areas like cybersecurity. Selected partners must adhere to strict protocols to obtain and maintain early access.
Sponsored Protocol
For more on related AI model restrictions, see the article on OpenAI Launches GPT-5.6 Sol, Terra and Luna in Limited Preview for Just 20 Partners After US Request. Additionally, according to authoritative sources like Wikipedia, OpenAI continues to innovate in artificial intelligence, balancing progress and responsibility.
