video input support

[zwukong]

qwen3 vl supports video understanding,

  # messages = [
    #     {
    #         "role": "user",
    #         "content": [
    #             {
    #                 "type": "video",
    #                 "video": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2-VL/space_woaudio.mp4",
    #             },
    #             {"type": "text", "text": "这段视频有多长"},
    #         ],
    #     }
    # ]

[basxto]

What would you run that with?

So far it looks like OpenAI API does not support video input. For Ollama I’m certain that it doesn’t (yet).

[zwukong]

anything by OpenAI API, ollama soon. There is a pull request