Python-Use implements a task-driven, result-oriented execution paradigm where LLMs interact with a complete Python interpreter in a continuous loop: Task, Plan, Code, Execute, Feedback. It provides models two core capabilities—API Calling through auto-generated Python code and Packages Calling via Python’s ecosystem—enabling autonomous understanding, planning, code writing, debugging, and execution. Users only provide task descriptions while the model handles all implementation internally, distinguishing this from code generators or IDEs.