Compare AI services using the same small task

Choose a task you understand well enough to judge. Give each service the same input and evaluate accuracy, missing details, useful output and the work needed to correct it.

Keep private information out of test prompts. Record the service and model when available because behavior and limits change.