OpenAI has announced a limited preview launch of its new GPT-5.6 model series, which aims to improve performance in the areas of programming, cybersecurity, and biology, while supporting complex, long-term tasks.
The company explained in a statement that the series includes three main models, led by the Sol model, confirming that it is equipped with a protection system that it described as the strongest to date, while enhancing its capabilities to discover security vulnerabilities and reduce misuse.
It added that it conducted extensive security tests that included about 700,000 processing hours, in cooperation with external parties that will continue to evaluate the model during the preview period.
OpenAI reported that the general availability of the GPT-5.6 series will begin in the coming weeks in coordination with US government agencies.
