On "Gemini Live audio"

Google's speech-to-speech models are real competition now, but the win here isn't the tech—it's that they're making it trivial to build on top of. A web UI you can prototype in minutes shifts the game from "which model is technically better" to "which one your users will actually adopt." The real pressure isn't on OpenAI's capabilities; it's on making these tools so frictionless that switching costs disappear. If developers start choosing based on ease and integration speed rather than raw performance, Google just reset the competitive table.