GPT-Realtime-Translate

Overview

GPT-Realtime-Translate is OpenAI‘s live speech-to-speech translation model, released to the API in May 2026 as part of a three-model real-time voice release alongside GPT-Realtime-2 and GPT-Realtime-Whisper.

Timeline

• 2026-05-11-AI-Digest — OpenAI releases GPT-Realtime-Translate to the API, priced at $0.034/minute. Supports 70+ input languages and 13 output languages for live speech-to-speech translation. Billing is by-the-minute (not by-the-token), reflecting a utility-bandwidth framing rather than the reasoning-density framing of GPT-Realtime-2. The per-minute rate for production-ready real-time translation materially lowers the floor for multilingual consumer apps and voice-first workflows that previously required stitching together STT + LLM + TTS pipelines.

Key Developments

1. Production-Ready Per-Minute Translation: $0.034/minute pricing for live speech-to-speech translation across 70+ input languages is the first generally available API model from a frontier lab at this price point for real-time conversational translation — lowers the economic floor for multilingual consumer products significantly.

Related

See also: OpenAI, GPT-Realtime-2, GPT-Realtime-Whisper, MOC - Developer Tools.