OpenAI's GPT-5.6 arrives with stronger coding and reasoning. Here's what changed
&imwidth=600&imheight=450&format=webp&quality=medium "OpenAI's GPT-5.6 arrives with stronger coding and reasoning. Here's what changed")
OpenAI's GPT-5.6 arrives with stronger coding and reasoning. Here's what changed
OpenAI has introduced the GPT-5.6 family of AI models, featuring improvements in reasoning, coding and cybersecurity. However, the company is starting with a limited preview after briefing the US government on the models' capabilities.
OpenAI has announced the GPT-5.6 family, introducing three new artificial intelligence models GPT-5.6 Sol, Terra and Luna. The new models are designed for different workloads, from advanced scientific research to everyday AI applications. Unlike previous launches, however, GPT-5.6 will initially be available only to a small group of trusted partners before a broader rollout in the coming weeks.
Three models for different tasks
GPT-5.6 Sol is OpenAI's flagship model, built for demanding reasoning, coding and scientific work. Terra is aimed at enterprise and developer use, offering performance similar to GPT-5.5 at roughly half the cost. Luna is the fastest and most affordable option, targeting applications where speed and lower costs matter most. OpenAI also says Sol introduces "Max Reasoning Effort", allowing the model to spend more time solving difficult problems, alongside "Ultra Mode", where multiple specialised AI agents collaborate on complex tasks.
Better coding and stronger security
According to OpenAI, GPT-5.6 delivers improvements in coding, planning and computational biology while using fewer computing resources than previous models. The company also highlighted stronger cybersecurity capabilities, saying the model performs well on vulnerability research benchmarks while remaining below its internal "Cyber Critical" risk threshold. To improve safety, OpenAI added expanded safeguards, including real-time monitoring, automated and human red-team testing, account-level abuse detection and stricter controls for cyber-related requests.
Why is GPT-5.6 restricted?
Trending Stories
&im=FitAndFill=(700,400))
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
)
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
The limited rollout follows consultations with the US government before launch. OpenAI said it agreed to begin with a controlled preview while US authorities develop a broader framework for evaluating advanced AI systems. The company stressed that the restriction is temporary and plans to expand GPT-5.6 across ChatGPT, the API and Codex over the coming weeks, adding that government preview requirements should not become the long-term approach for releasing frontier AI models.
