AI Models & Assistants / Knowledge library
Agent Turn
Processes current context through a model and handles the resulting action or answer
Why learn Agent Turn?
It provides the repeatable step from which larger agent loops are constructed.
The problem it solves
It converts evolving run state into one controlled model decision at a time.
Where it came from
A turn commonly includes context assembly, a model response, and classification of that response as a tool request, handoff, or final output. Tool results are appended before the next turn.
Build your understanding
Work through these exercises in order. Try the task before opening a reference.
- Define Agent Turn and identify its place in an agent architecture.
- Choose a task where Agent Turn improves an agent system.
- Diagnose one failure caused by misusing Agent Turn.
- Interpret a trace or state transition involving Agent Turn.
- Explain one operational tradeoff of Agent Turn.
Documentation and sources
- OpenAI Agents SDK: Running agents (openai.github.io)