ByteDance's SeedRealtime watches, listens and talks — all at the same time
ByteDance's Seed team released SeedRealtime on Wednesday, a model that natively fuses audio, video and text in one end-to-end system for full-duplex interaction: it keeps watching and listening even while it speaks, instead of taking turns. The company says it reads sound, visuals and timing together to work out who is addressing it and what they want, and it is already live in the Doubao assistant app. It extends April's voice-only Seeduplex model to the visual world.
Why it mattersFull-duplex voice was step one; adding eyes makes an AI assistant behave less like a walkie-talkie and more like someone in the room. ByteDance shipping this straight into its flagship consumer assistant signals that real-time multimodal interaction is becoming the next battleground for AI apps.
Read in the app — free, in 9 languages
Related stories
Apple Music will tell you when a song was made by a machine — but the uploader decides whether to say so2026-08-21DeepSeek's cheap workhorse can now see — and on agent tasks that need eyes it says it is close to Anthropic's best
2026-08-21Adobe will now generate the music, the voiceover and the door slam — and it says the licence covers you
2026-08-21Show the robot once — three to twelve seconds — and it gets the job right 59 times out of 100 with no training at all
2026-08-21Six months ago Hollywood sent ByteDance a cease-and-desist. This week they signed a truce — and no money changed hands.
2026-08-20