I tried testing it, but reasoning effort parameter doesn't seem to be there and output is sort of broken because it reasons directly in the output tokens...