OpenAI: gpt-realtime API
OpenAI's flagship multimodal realtime model, enabling end-to-end, low-latency interactions across text and audio. It natively understands nuances like inflection, emotion, and pace for human-like conversations.
- Context window: 32,000 tokens
- Max output: 4,096 tokens
- Input: text, audio
- Output: text, audio
- Tool calling: Supported
- Released: 2025-08-27
- Knowledge cutoff: 2023-10-29
Frequently Asked Questions
What is the context window of gpt-realtime?
gpt-realtime supports a context window of up to 32,000 tokens.
Does gpt-realtime support function calling?
Yes. gpt-realtime supports tool / function calling.
What is the knowledge cutoff of gpt-realtime?
The knowledge cutoff of gpt-realtime is 2023-10-29.