OpenAI unveiled its latest family of models on Thursday, adding three new options to an increasingly competitive AI market. The GPT-5.6 lineup includes Sol, described as the company's workhorse model; Terra, a mid-range option; and Luna, the budget-friendly entry. The company says these models deliver significant improvements across enterprise tasks, coding, and scientific research, promising more power at lower cost than previous generations.

CEO Sam Altman told CNBC that Sol is 54% more token-efficient for AI coding work, a key metric for developers watching their compute budgets. OpenAI also highlighted the models' cybersecurity strengths, calling GPT-5.6 its strongest offering yet for threat modeling, code review, and blue-team simulations — defensive exercises where organizations test their own systems. The company released a new workplace tool called ChatGPT Work, designed to help enterprise teams with drafting documents, building spreadsheets, and creating presentations across desktop, web, and mobile.

The launch comes days after similar announcements from Meta and a rival AI startup, but OpenAI's messaging clearly targets Anthropic, which has gained traction by focusing on enterprise clients. OpenAI points to the Artificial Analysis Coding Agent Index, claiming Sol outperforms Anthropic's top model by 2.8 points on the benchmark while using fewer tokens and costing about a third less. The company says Terra and Luna also beat their direct competitors. GPT-5.6 is now available via ChatGPT, Codex, and the OpenAI API, with per-million-token pricing ranging from $1 input and $6 output for Luna up to $5 input and $30 output for Sol.

Source: TechCrunch