Tencent’s HunyuanVideo-Foley, Microsoft’s MAI Models, and OpenAI’s gpt-realtime API

In today’s AI Deep Dive, we explore major AI breakthroughs reshaping voice, translation, and media. Microsoft debuts its first in-house AI models, including MAI-Voice-1 for expressive speech and MAI-1-preview, a versatile foundation model. OpenAI rolls out gpt-realtime, a speech-to-speech model with enhanced reasoning and production-ready API features for next-gen voice agents. Meanwhile, Command A Translate emerges as a secure, high-quality enterprise translation solution, and Tencent open-sources HunyuanVideo-Foley, bringing synchronized, professional-grade audio to AI video production.