Hacker News Reader

We've been running a small finetuned VLM for OCR and yeah... cost is like 1/3rd of the usual vision APIs. Small models are kinda the whole product for us

View on HN

Which vlm are you using? Ive found a reddit thread listing like 20 of them, but its hard to find any concrete info on what is worth even tryin.

Reply on news.ycombinator.com