Qwen3-VL can scan two-hour videos and pinpoint nearly every detail

[brennan]

Qwen3 models are really good. We have been using them on POCs lately.