Microsoft · 2026-08-02 · rumor
MAI-Realtime rumor — Microsoft's full-duplex voice model hidden in MAI Playground
MAI-Realtime is a bidirectional voice model Microsoft is quietly testing in its MAI Playground — the first native full-duplex speech model from the in-house MAI family, listening and speaking at the same time.
TestingCatalog spotted an unannounced Microsoft voice model — MAI-Realtime — running as a hidden preview alongside the public MAI models.
What is it?
MAI-Realtime is an unannounced bidirectional voice model from Microsoft's in-house MAI team, discovered by TestingCatalog as a hidden entry in the MAI Playground on August 2, 2026. Unlike Microsoft's existing MAI-Voice-1 (text-to-speech) and MAI-Transcribe-1.5 (speech-to-text), MAI-Realtime is a single native speech-to-speech model that listens and speaks at the same time.
How does it work?
The MAI-Realtime preview panel exposes 17 languages — English, German, Spanish, French, Italian, Portuguese, Japanese, Korean, Chinese, Dutch, Hindi, Indonesian, Arabic, Russian, Turkish, Vietnamese, Thai — and two voices, Victoria and Grant. Turn-taking is configurable: a Switchboard mode uses the MAI-Ears endpointer while a deterministic mode relies on silence detection plus a Whisper semantic endpointer. TestingCatalog reports the model can call web search and other tools mid-conversation.
Why does it matter?
A native full-duplex model would give Microsoft its own answer to OpenAI's realtime voice API and Sesame — the last piece Copilot needs to run voice agents end-to-end without leaning on OpenAI infrastructure. Microsoft has not published a model card, pricing, launch date, or added MAI-Realtime to the public MAI catalogue; TestingCatalog frames it as a tightly controlled partner preview, not an imminent public launch.
Who is it for?
Microsoft-ecosystem developers watching Copilot's move away from OpenAI voice