Model
151d agoQwen3.5‑Omni Learns Code From Speech
Alibaba’s new Qwen3.5‑Omni model can understand text, images, audio, and video, surpassing rivals on audio tasks and unexpectedly learning to generate code from spoken instructions and visual input.
The Signal
This multimodal breakthrough shows how models can acquire new skills without explicit training, hinting at a future where AI systems adapt fluidly across media and domains worldwide.