1.5 added support for step-1.5v-mini and gemini 2.0 thinking
This is the second native video inference API model.
You can visit their official website here.
Additionally, waiting time, retry times, and long video segmentation time have been added to the script.
Now we can customize the video segmentation interval.
Of course, for alignment considerations, it is still recommended to use the default 5 minutes.