Qwen3.5-Omni was launched as we speak (March 30, 2026) by Alibaba’s Tongyi Lab. This omnimodal mannequin can perceive textual content, photographs, audio, and video, and generate each textual content and speech. Key options: Thinker-Talker structure with Hybrid-Consideration Combination of Consultants, 256K token context, 100M+ hours of multimodal coaching, 113 language speech recognition, ARIA know-how for text-speech alignment, and Audio-Visible Vibe Coding (watch movies and write practical code). Surpasses Gemini 3.1 Professional in audio/video understanding and beats ElevenLabs/GPT-Audio on voice benchmarks. Entry through DashScope API or HuggingFace Transformers (80GB VRAM for full mannequin).
Proceed studying
Audio-Visible Vibe Coding with Qwen3.5-Omni: Write Code from Video Alone
on SitePoint.
