mixture-of-experts

Qwen3.5-Omni: New Multimodal AI Model with 256K Context

Qwen3.5-Omni scales to hundreds of billions of parameters with 256K context length, achieving SOTA results across 215 audio-visual benchmarks and introducing Audio-Visual Vibe Coding capability.