Microsoft Unveils Its First Speech-to-Speech Model MAI Realtime 1: It Can Be Interrupted, Make Calls and Get Work Done, at Only One-Third the Price of Competitors
Microsoft launched its first speech-to-speech model, MAI Realtime 1, at Ignite 2026, targeting real-world enterprise scenarios. It can handle conversational nuances such as interruptions, quick acknowledgments, and waiting for the other party to finish speaking; it can also understand requests in noisy environments or when multiple people are speaking at once. When deep thinking is required, it can call on a more powerful reasoning engine, connect to enterprise systems to complete operations such as bookings and customer service, and its tone and style can be tuned with prompts. Price is the big focus: basic speech-to-speech usage is billed by audio, with aggressive pricing.