China will always stay one step ahead of the U.S. in AI Alibaba just dropped Qwen3-Omni and it's the first model that doesn't sacrifice anything to be multimodal. Usually when you add vision and audio to a language model, text performance tanks. Not here. Qwen3-Omni-30B-A3B