We don't share documentation on how our AI is trained. We use different models, instructions, and training data depending on the tool to deliver what we feel is the best result for each use case.
