“Qwen VLo is still in the preview stage, and there are many shortcomings.”
Alibaba’s new image model understands and generates pictures in one system. The 3,250-word launch post has no benchmarks, no parameter count, no weights and no API. Its own limitations list includes inaccuracies, ignored instructions and instability. It is a preview of a preview.