GPT-Realtime-2 at a glance
GPT-Realtime-2 added configurable reasoning to speech-to-speech agents, alongside new streaming translation and transcription models.
What changed with GPT-Realtime-2
On May 7, 2026, OpenAI released GPT-Realtime-2, a realtime voice model with configurable reasoning for speech-to-speech agents. It shipped with GPT-Realtime-Translate for streaming speech translation and GPT-Realtime-Whisper for streaming speech-to-text.
The model page highlights stronger instruction following and more reliable tool use in complex voice workflows. It accepts text, audio and image input, has a 128K-token context window, and runs on the Realtime endpoint. Audio is priced at $32 per million input tokens and $64 per million output tokens.
GPT-Realtime-2 benchmarks and pricing
Values are copied from the linked source on the date shown. They are not IA Compa measurements.
- Audio input price per 1M tokensas of · Source for Audio input price per 1M tokens: developers.openai.com (opens in a new tab)
- $32.00
- Audio output price per 1M tokensas of · Source for Audio output price per 1M tokens: developers.openai.com (opens in a new tab)
- $64.00
- Context windowas of · Source for Context window: developers.openai.com (opens in a new tab)
- 128K tokens
Where to compare GPT-Realtime-2
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
Leaderboards
General leaderboards
Why GPT-Realtime-2 matters for businesses
For phone agents that book appointments, qualify leads or answer FAQs, better tool reliability means fewer dropped tasks mid-call. Higher reasoning effort can add latency, so teams should tune it per use case. OpenAI later added gpt-realtime-2.1 and its mini version in July 2026.
Put it to work: AI Voice Agents
Sources for this story
- OpenAI API changelog (opens in a new tab)Official
OpenAI · developers.openai.com
- GPT-Realtime-2 model page (opens in a new tab)Official
OpenAI · developers.openai.com
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #voice
- #realtime-api
- #openai
- #speech-to-speech
