Launching today
Benchpress
Every OCR model is #1 somewhere. Test them on your own pages
10 followers
Every OCR model is #1 somewhere. Test them on your own pages
10 followers
Every OCR model is topping some benchmarks. But can they lift your documents? 💪 Drop in one page, and every major OCR and document AI reads it. See who got it right, where each one slipped, and who was fastest and cheapest.
Hi PH 👋 Arindam from Rasyn.
Every OCR model is #1 on some leaderboard, usually one its own company published. We got tired of it. None of those benchmarks used your documents.
So we built Benchpress. To see if these models can even lift? (Your Documents) 💪🏻
Drop in one page. Every major API reads it. You see how good they do on your documents, and where each one slipped.
We are on the list too, so first or last does not matter to us; transparency and truth matter.
Don't take our word for it. Try your hardest page.
@arindam9 Hey everyone, Inderpreet here, the other co-founder 👋 Benchpress started as an internal tool. While building our own document pipeline, we kept picking models off leaderboards, and they didn't work well on our documents. So we started testing them ourselves, and that's how we figured out which model actually works for which kind of page.
Would love to hear how it does on your documents. The more kinds of pages get tested, the clearer it gets which models work where, and hopefully that helps push OCR to get better too.