[Info] Models that natively support audio and video

[gzhihongwei]

Hi there,

Are there any other models that natively support audio and video besides the Gemini API? I'm aware that VideoMME uses captions as a means of assessing the models that support video and text, but seems like there is no model besides Gemini Pro that supports audio and video natively? If there aren't any more, are there any models that you suggest benchmarking (i.e. besides the ones in VideoMME)?

Thanks,
George

[gzhihongwei]

Also, is there any plan to add support for models that support audio and video?

[kcz358]

There will be such plan in the future but it will be added slowly. I think the most recent progress is that #461 that supports the first video+audio benchmark and model (Gemini)