Can small AI models think as well as large ones?

In the absence of a testable definition or “thought”, I will make no judgment.