Duo Unveils Reciprocal World Model for Full‑Duplex Interaction
| Source: HF Papers | Original article
Ant Group has unveiled **Duo v1**, the first implementation of a “Reciprocal World Model” (RWM) that supports full‑duplex interaction. The system continuously ingests audiovisual streams while simultaneously generating real‑time responses, blurring the line between perception and action. The research paper – authored by Hengyuan Cao, Biao Gong and Guoshan Lu – frames the RWM as a world where intelligence and physical reality are co‑fundamental, a departure from conventional models that treat perception as a one‑way feed‑forward process.
The launch matters because it demonstrates a concrete step toward AI agents that can engage users as fluidly as a human conversation partner, reacting to visual and auditory cues without pause. By handling input and output in parallel, Duo v1 promises more natural, interactive experiences for applications ranging from virtual assistants to immersive simulations. The open‑source repository on GitHub already provides the codebase, inviting the research community to explore extensions and benchmark the approach against existing multimodal models.
Looking ahead, the community will be watching for performance metrics, especially latency and fidelity in real‑time settings, as well as any follow‑up releases that expand the model’s capabilities or scale. Integration into downstream products, collaborations with hardware vendors, and comparative studies with other interactive world models – such as the BirdDreams BEV video generator we covered earlier – will shape how quickly full‑duplex AI becomes a mainstream tool.
Sources
Back to AIPULSEN