They should've tested other embedding models, there are better ones than openai'... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		383toast 63 days ago \| parent \| context \| favorite \| on: Production RAG: what I learned from processing 5M+... They should've tested other embedding models, there are better ones than openai's (and cheaper)

prettyblocks 63 days ago [–]

Which do you suggest?

roze_sha 63 days ago | | [–]

https://huggingface.co/spaces/mteb/leaderboard

remz14 63 days ago | | | [–]

You should use RTEB instead. See here for why: https://huggingface.co/blog/rteb

Here is that leaderboard https://huggingface.co/spaces/mteb/leaderboard?benchmark_nam...

Voyage-3-large seems like SOTA right now

383toast 63 days ago | | | | [–]

yep

leftnode 63 days ago | | [–]

The Qwen3 600M and 4B embedding models are near state of the art and aren't too computationally intensive.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact