ByteDance launches SeedRealtime for real-time audio-visual interactions, intensifying the AI race

ByteDance just dropped a new AI model that can see you, hear you, and talk back to you, all at the same time and in real time. The company’s internal AI research group, Seed, launched SeedRealtime on April 9, making it available to users of the Doubao app, which has quietly become one of the most widely used consumer AI products on the planet. SeedRealtime is what’s called a “full-duplex multimodal model.” In English: it can process audio, video, and text simultaneously while you’re still talking, rather than waiting for you to finish before responding. What SeedRealtime actually does The model reportedly delivers a 12% improvement in conversational fluency compared to its predecessors. ByteDance attributes this to advanced listening techniques and interference suppression, which essentially means the AI got better at filtering out background noise and not tripping over itself mid-sentence. The model is being deployed on Doubao, ByteDance’s flagship AI assistant app. The app boasts 155 million weekly active users, with daily active users surpassing 200 million shortly after the Lunar New Year in 2026. Doubao powers interactions across multiple ByteDance products, including Coze, Ji...