Helo Dejan! Just throwing it out there…. I’m guessing you have, but have you tried Built-in AI APIs?
https://developer.chrome.com/docs/ai/built-in-apis#api_status

Hi Dejan! Which model did you use for the embeddings? I understand that both accuracy and speed depend on the model used, some models lose less quality when quantized.

Another question: any experience with non-English languages? There are more models out there now, but it’s tough to find free ones that match the performance of the English ones.

Thanks!