Alibaba’s new Qwen3‑VL can scan two‑hour videos, crushing benchmarks with 96.5% on DocVQA and 875 on OCRBench. Could this be the GPT‑5 of vision‑language? Dive into the open‑source breakthrough! #Qwen3VL #DocVQA #OCRBench

🔗 aidailypost.com/news/qwen3vl...