Hi there - we recently fixed this issue and will release a new version for it soon!
In-browser unified multimodal understanding and generation.