# OpenAI has introduced three new audio models through its API.

*genai · news · 2026-05-07 · Interesting Engineering*

## Key points

- OpenAI released GPT-Realtime-2 with GPT-5-class reasoning for real-time voice interactions.
- GPT-Realtime-2 expands its context window from 32K to 128K for longer, more complex conversations.
- The GPT-Realtime-Translate model supports real-time translation of over 70 input languages to 13 output languages.
- Deutsche Telekom and Zillow are already building live customer support and voice assistant tools using these models.
- GPT-Realtime-2 showed a 15.2% performance improvement on Big Bench Audio benchmarks over its previous version.

**Companies:** OpenAI, Deutsche Telekom, Zillow
**Countries:** United States, Germany

[Read the full story on Interesting Engineering](https://interestingengineering.com/ai-robotics/openai-gpt-realtime-2-voice-ai-models)

---

Canonical: https://newsio.io/n/95a9dd6b-0786-467f-8d73-8a2eab34ce91/openai-has-introduced-three-new-audio-models-through-its-api-the-new-models-aim
Summarized by Newsio from Interesting Engineering. https://newsio.io/how-it-works