ByteDance Launches SeedRealtime: A Native Audio-Visual Full-Duplex LLM
iamrobotbear · x · 2026-08-06
ByteDance has launched SeedRealtime, a native audio-visual full-duplex LLM. Using a unified architecture, the model natively fuses audio, video, and text to enable real-time interaction over continuous multimodal streams, delivering a "watch, listen, and speak" experience.
It is now live and free on the Doubao App, supporting real-time voice and video conversations. In complex scenarios like group dinners, the model can accurately map names to faces after a brief introduction and consistently distinguish between multiple speakers during chaotic, overlapping conversations.
Related event: ByteDance Launches Full-Duplex SeedRealtime Model on Doubao(7 posts)→
More from Models
- Frontier Models Tested on Complex Agents: DeepSeek Wins on Cost Despite Inefficiency — rohanpaul_ai · 2026-08-06
- Mistral's Voxtral TTS Hits 70ms Latency but Stays Closed Source — shashib · 2026-08-06
- Top AI Models Score Under 50% on New Math Figure Reasoning Benchmark — prof_g · 2026-08-06
- Ethan Mollick: LLMs Improve at Following Instructions but Exercise More 'Judgement' — emollick · 2026-08-06
- Frontier LLMs Perform Best in Week 1: Dev Calls for a Proof of Model Standard — sull · 2026-08-06
- Polymarket Favorites Anthropic at 69% to Win AI Race by 2026 — Polymarket · 2026-08-06