AI summarized from verified sources
OpenAI Launches GPT-Realtime-2 for Advanced Voice Agents (48 chars)
Build advanced real-time voice agents to boost productivity (52 chars).
SOURCE CHECK
3 sources
Sources
Key Points
- 1128K context for long talks.
- 2Improved tool calls & recovery.
- 370-lang translation & streaming.
- 4Test in Playground now.
OpenAI released GPT-Realtime-2, Translate, and Whisper in Realtime API. GPT-Realtime-2 handles 128K context and tool calls in real-time for complex voice tasks. Developers can build voice agents easily, with low-latency translation and transcription. Pricing starts at $32/1M input tokens.
Key points
OpenAI strengthened the voice API beyond simple speech features, adding models that can reason, translate, and transcribe while a conversation continues. The result is a more fluid experience.
Impact
This is useful for call centers, learning tools, and automatic meeting notes—any product that starts with voice. It is a strong fit for developers who want interactions that feel more natural than typing.
What changed
OpenAI introduced a set of voice models in the API that can reason, translate, and transcribe while people speak. Developers can build more natural support agents and multilingual assistants. The update focuses on both quality and low-latency voice experiences.
Briefs that include this news
Use daily, weekly, and monthly briefs to understand the surrounding context.