Lanthanum.AI is a project with which the user can access small LLM models through web which would be running via WEB GPU and this doesn't require any kind of AI , Everything runs in local and nothing goes out of users pc.

Then during hackbery I added More video captioning realted feature which you can see below

• Added Live Video captioning feature (aka VLM) it uses Apple™ FastVLM 0.5B Model to caption the images and give it respones , The respones can be customized too
• Added COLAB support https://colab.research.google.com/drive/1-fcmgaB8ztEXFp-RlFqZDJ8aep8G7eGc?usp=sharing

git clone https://github.com/Enderman8843/WEB_AI
cd WEB_AI
npm run dev
v

https://lanthanum.vercel.app/

• Lanthanum.AI is best recommened to run on Chromium based browser i.e Chrome and Edge >= 113 ,3 ,
• If using other broswers check this guide https://caniuse.com/webgpu
• This dosent work if your pc dosent support DirectX12

If you can't use webGPU models try using cpu mode(unstable)