Real-time multimodal models for interactive AI characters and worlds
Vivix builds real-time interactive multimodal models for AI characters, scenarios and worlds that users can steer through text, images, video, voice, touch and motion. Its model lineup includes Vivix-A1, a full-duplex model for expressive full-body AI characters, and Vivix-W1, a streaming-native model for continuously generated audiovisual worlds, alongside Vivix IVI, a real-time video runtime for interactive stories. Developers can access the models through the Vivix API platform. Vivix-W1 was launched publicly in October 2026 with per-second usage pricing.