Alibaba Technical Report: Qwen3-VL beats GPT-5 and Gemini 2.5 Pro on visual tasks and has 100% accuracy on “needle-in-a-haystack” tests for 30-minute videos — A few months after launching Qwen3-VL, Alibaba has released a detailed technical report on the open multimodal model.
Posted from: this blog via Microsoft Power Automate.
