OpenAI launches GPT-Realtime-2 and translation features in its API
openai added three new voice models to its realtime api: gpt-realtime-2 features gpt-5-class reasoning, gpt-realtime-translate supports over 70 input languages, and gpt-realtime-whisper offers live speech-to-text. the models target customer service and education use cases and include guardrails against abuse and fraud.
OpenAI expanded its Realtime API with three new voice intelligence models. GPT-Realtime-2 utilizes GPT-5-class reasoning to handle complex user requests, while GPT-Realtime-Translate supports over 70 input and 13 output languages. A third model, GPT-Realtime-Whisper, provides live speech-to-text transcription. These tools aim to shift interfaces from simple call-and-response to systems that can reason and act during conversations. The company implemented guardrails to prevent spam, fraud, and policy violations. Billing varies by model, with Translate and Whisper charged by the minute and GPT-Realtime-2 by token consumption.