•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
OpenAI has announced three new voice features in its API suite aimed at developers building applications for real-time conversation, translation, and transcription.
The latest model, GPT-Realtime-2, is designed to generate natural-sounding speech and interact with users directly. OpenAI says it includes reasoning capabilities comparable to GPT-5, enabling it to handle more complex requests than models that primarily follow pre-scripted prompts.
GPT-Realtime-Translate is a real-time translation feature intended to match the pace of natural conversation. It supports more than 70 input languages and 13 output languages.
GPT-Realtime-Whisper provides live transcription during a conversation, producing speech-to-text as the dialogue unfolds.
OpenAI positions the new models as a step beyond basic question-and-answer voice interactions, toward a more complete voice interface that can listen, reason, translate, transcribe, and act throughout a conversation.
OpenAI says businesses looking to expand customer-service capabilities are among the most obvious target users. It also notes the features are suitable for education, media, live events, and content-creation platforms.
On misuse risk, OpenAI says safeguards are in place to prevent abuse such as spam, scams, or other forms of online abuse. It also states the system can automatically disconnect a conversation when policy violations are detected.
All three models are part of OpenAI’s Realtime API. OpenAI says GPT-Realtime-Translate and GPT-Realtime-Whisper are priced per minute, while GPT-Realtime-2 is priced per token.
Premium gym chains are entering a “golden era” that is ending or already in decline, as rising operating costs collide with shifting consumer preferences toward more flexible, community-based ways to exercise. Long-term memberships are shrinking, margins are pressured by higher rents and facility expenses, and competition from smaller, more personalized…
