End-to-End Omni (text, audio, image, video, and natural speech interaction) model based Qwen2.5
Generate text and speech responses from various inputs