Real-time, full-duplex model for expressive full-body AI characters
Vivix-A1 is Vivix's real-time, full-duplex model for expressive, full-body AI characters that listen, respond, move and act inside open worlds. One model handles listening, speaking and movement at once for low-latency conversation with gestures, micro-expressions and interruption handling, and characters can hold and demo products live. It is available through the Vivix API as a streaming avatar model (vivix-a1-stream) and for script-, audio- and image-driven avatar video generation (vivix-a1).