OpenAI has started previewing GPT-5.6 Sol, and the model family now has a clearer shape: Sol, Terra and Luna are the three durable tiers.
The company is using a limited preview first. GPT-5.6 models will initially be available through the API and Codex to a select group of trusted partners and organizations. OpenAI says broader access for people using ChatGPT, Codex and the API is planned soon.
The naming is also new. The number marks the model generation, while Sol, Terra and Luna mark capability tiers. Sol is the flagship model. Terra is the balanced model for everyday work. Luna is the fast and more affordable option.
Sol, Terra and Luna are the story
The official image on OpenAI’s news page makes the tiering easy to understand. Sol is represented by the sun, Terra by Earth and Luna by the moon. It is a cleaner way to explain the family than a pile of model suffixes.
OpenAI says Terra has competitive performance to GPT-5.5 while being twice as cheap. Luna is meant to bring strong capability at the lowest cost in the family. That puts Sol at the top for the hardest work, Terra in the practical middle and Luna in the high-volume lane.
Tech My Money recently covered reports that OpenAI would release ChatGPT 5.6 to government-approved customers first. The official preview now gives that story a clearer frame. OpenAI says it shared the model plans with the U.S. government and is starting with trusted partners before a broader rollout.
The benchmarks are part of the announcement
OpenAI is not treating GPT-5.6 as a simple chatbot refresh. The company says GPT-5.6 Sol is its strongest model yet, with gains in coding, biology and cybersecurity. It also introduced a higher reasoning effort and an ultra mode that can use subagents for complex work.
For coding workflows, OpenAI says GPT-5.6 Sol sets a new state of the art on TerminalBench 2.1. The chart places GPT-5.6 Sol Ultra and GPT-5.6 Sol ahead of Claude Mythos 5, GPT-5.6 Terra, Claude Fable 5, GPT-5.5, GPT-5.6 Luna and other models in that test.
OpenAI also points to biology workflows. Its GeneBench chart shows Sol, Terra and Luna scaling differently by output tokens, with Sol reaching higher scores while using fewer tokens than GPT-5.5 in OpenAI’s summary.
Safety is central to the preview
The cybersecurity section matters because OpenAI is pairing the stronger model with extra safeguards. The company says GPT-5.6 Sol launches with its most robust safety stack to date, including stronger protections for higher-risk activity, sensitive cyber requests and repeated misuse.
OpenAI says Sol, Terra and Luna all show stronger cyber capability as reasoning increases. The company also says it spent weeks pressure-testing the system, running automated red-teaming and using human expert red teams during the preview period.
Pricing starts at $1 per million input tokens
OpenAI lists GPT-5.6 pricing per 1 million tokens. Sol costs $5 for input and $30 for output. Terra costs $2.50 for input and $15 for output. Luna costs $1 for input and $6 for output.
The company is also changing prompt caching. GPT-5.6 supports explicit cache breakpoints and a 30-minute minimum cache life. Cache writes cost 1.25 times the uncached input rate, while cache reads keep the 90% cached-input discount.
For now, the practical takeaway is simple. GPT-5.6 is real, the model family is no longer just one name, and OpenAI is using the preview period to test capability, safety and pricing before wider access.
