Live drop
XGEN LabsDropped Sep 17, 20261 build1 maker
Egocentric first-person video+audio generation conditioned on actions and observation history for persistent world navigation and interacti….
About this release
XGEN-JING is an egocentric interactive experience model (JING-Flash-v1) that generates first-person video and synchronized audio for navigation, object manipulation, and dialogue. It is conditioned on actions, reference images, and observation history, and is released with weights and inference code on Hugging Face. It forms one half of XGEN Labs' Generative World Simulation system alongside the DAO engine, enabling agentic first-person experiences in evolving shared worlds. It ranked #1 on WBench Full split as of September 17, 2026.
Read the official XGEN Labs launch pageWorth opening first
Paste the URL and Dropday.ai will credit you, track the build as it evolves, and put it on this drop.
Submit a build