OpenAI has released a new generation of artificial intelligence, GPT-5.6, and given its models distinct names: Sol, Terra, and Luna. The launch was reported across multiple outlets including Yahoo Tech and CCN.com.
The standout among them appears to be Sol. According to SecurityWeek, OpenAI is positioning GPT-5.6 Sol as its most advanced cybersecurity AI to date. Coverage from Il Sole 24 ORE frames the broader release as aimed at coding and cybersecurity tasks, two areas where AI assistants are increasingly used by professional developers and security teams.
The move is being read as a direct competitive play. Il Sole 24 ORE describes the launch as OpenAI renewing its challenge to Anthropic, the rival AI lab. Both companies have been racing to build models that can write, review, and secure software code.
The new models also emphasize efficiency. According to Basic Tutorials, GPT-5.6 delivers more performance while using significantly fewer tokens. Tokens are the units of text an AI processes, and they directly affect how much a model costs to run, so fewer tokens for the same or better results can mean lower expenses for the businesses and developers building on top of OpenAI's technology.
Beyond the naming and the efficiency claims, the sources provide limited technical detail about how Sol, Terra, and Luna differ from one another or how they compare against earlier versions.
Why it matters: AI is moving deeper into specialized, high-stakes work like cybersecurity and software development, and OpenAI's push to make these models cheaper to run signals an intensifying battle with rivals like Anthropic over who powers the next wave of professional AI tools.