Introducing gpt-realtime and Realtime API updates for production voice agents

[proflead]

@edwinarbus

The Realtime API is officially out of beta and ready for your production voice agents!

We’re also introducing gpt-realtime—our most advanced speech-to-speech model yet—plus new voices and API capabilities:

Remote MCPs
Image input
SIP phone calling
Reusable prompts

gpt-realtime was trained with customers to excel at real-world tasks like support, personal assistance, and education. It’s better at:

• Following instructions
• Calling tools
• Natural, expressive speech
• Understanding cues (like laughs)
• Switching languages

We’ve added two API voices, Cedar and Marin, and improved voice quality overall for more humanlike speech that can adapt to tone.

And we’ve cut pricing by 20%.

Prompting guide on speed, tone, hand-offs, and more: Realtime Prompting Guide

Docs: Realtime and audio | OpenAI API

SIP calling, reusable prompts, and remote MCP support make this release especially notable for moving voice agents from demos into operational workflows. I should try it in some project!