Official documentation
Voice agents
OpenAI, 2026
Read the originalRead on October 17, 2026
- What it is
- OpenAI developer guide that compares two ways to build a voice agent: a single speech-to-speech model through the Realtime API, and a chained pipeline of speech-to-text, an agent workflow and text-to-speech.
- How much weight it carries
- Official vendor documentation, so it describes OpenAI products and recommends its own APIs. Model names in the guide change often and were not all consistent when checked.
Facts taken from this source
- OpenAI says the chained path is the one to use when you want to inspect or change the text between speech recognition, the agent and speech generation.
“Use the chained path when you want to inspect or transform text between speech recognition, your agent, and speech generation.” (Build a chained voice workflow)
- OpenAI describes its speech-to-speech path as a single model that takes audio in, decides what to do and speaks back.
“Use one model to interpret audio, decide what to do, and respond in speech.” (Comparison table)