Sample a video into message content for createChatCompletion(). Each frame is an image part that follows its timestamp.
The model must support image input. It reads the frames as separate images, so a model with its own video layout (for example Qwen3-VL) uses more tokens than with sampleVideo().
Sample a video into message content for createChatCompletion(). Each frame is an image part that follows its timestamp.
The model must support image input. It reads the frames as separate images, so a model with its own video layout (for example Qwen3-VL) uses more tokens than with sampleVideo().