We leverage a number of the latest versions of frontier 4o models and we layer in more use-case specific models whenever possible.
Our team is constantly evaluating security, performance, and capabilities to make sure our AI steps are only getting better over time. You never have to worry about the specific models we’re utilizing, because we’re constantly evaluating and improving on your behalf.
With that being said, you can always break out into our API steps if you want to be very specific about model choice.
You can read more about AI Step Security here! Last modified on February 24, 2026