It would likely fall under the same policy that causes latest Opus/Fable to fallback to less capable models when working on ML code