This course is free — log in to save your progress and unlock the full library.

Gemini Embedding 2 goes multimodal — no more frame-by-frame video hacks

0:24

Gemini Embedding 2 supports multimodal embedding — huge. Until now I converted everything to text first, going through video frame by frame. That whole pipeline just collapsed into one API.

Transcript

Google has finally launched Gemini Embedding 2, which supports multi-model embedding. This is very huge because for my product, this one, I had to integrate all integration one by one. What do I mean by that? I had to convert everything to text, whether it is a video, I need to go through frame by frame, it's costing a lot. And then you have to embed all of that Gemini Embedding model. Comment link, I'll share you one of the projects which I'm working on.

←Previous1.52 Forward a reel, get it recreated from your own footage — building QEdit overnight Next1.54 I was burning lakhs on AI — open-source models on Modal do the same job→