ByteDance / Volcano Engine
The ByteDance Seed models page lists Seeduplex as a native full-duplex speech LLM with high-precision interference suppression and adaptive endpoint detection. It sits beside Seed LiveInterpret 2.0, Seed Realtime Voice and Seed-Music in the Seed speech and GenMedia portfolio. This profile tracks Seeduplex as ByteDance's English-facing voice-agent and real-time speech model signal.
Editorial verdict
Teams tracking Chinese full-duplex speech models, realtime voice agents and multimodal interaction research.
Avoid selecting it for production voice agents until API access, latency, languages and data terms are confirmed.
Seeduplex gives ByteDance Seed a distinct voice-interaction profile beyond text, image and video models.
Voice model access and pricing should be verified through BytePlus or Volcano Engine
BytePlus billing, Volcano Engine billing
Commercial use should follow the current product, API, model license and billing terms.
Review prompt, file, media upload, retention and training-use terms before sensitive workloads.
Track it for full-duplex speech conversations where interruption handling and endpoint detection matter.
Compare with MiniMax Audio, Qwen Audio, MiMo Speech and GLM Audio before choosing a hosted voice stack.
Model names, quotas, release status, regional access and commercial terms can change quickly; recheck official sources before procurement or production use.
minimax-audio
qwen-audio
mimo-speech
zhipu-glm-audio
official · en · verified 2026-08-04
Lists Seeduplex, Seed LiveInterpret 2.0, Seed Realtime Voice and Seed-Music.
official · en · verified 2026-05-17
Official Seeduplex model page linked from the Seed portfolio.
Last checked: 2026-08-04