New
GPT Realtime · 1M tokens
Low-latency speech-to-speech sessions for live voice agents with function calling.
What you get
- 1M tokens for every unit you buy. Order several units to stack quota.
- A personal access key for
gpt-realtime, issued automatically once your order is completed. - No shipping: delivery is digital, under My Account > Order History and by email.
Capabilities
- Realtime
- Voice
- Tools
Best for
Live voice agents, phone bots.
Quick start
from openai import OpenAI
client = OpenAI(api_key="YOUR_ACCESS_KEY", base_url="YOUR_BASE_URL")
# Realtime sessions use WebSocket/WebRTC with model="gpt-realtime"
| Model Specs | |
| Model ID | gpt-realtime |
| Context Window | 32K tokens |
| Max Output | 4K tokens |
| Modalities | Audio, text in / audio, text out |
| Capabilities | Realtime, Voice, Tools |
| Best For | Live voice agents, phone bots |
| Unit Size | 1M tokens |
$24.00
10 or more $22.80
50 or more $21.60
Virtual product
Instant digital delivery · No shipping
- API model:
gpt-realtime - Each unit: 1M tokens
Your access key is issued automatically once payment is completed and appears under My Account > Order History.