OCR Arena
AI
ocr model comparison
SHOW HN2025
How HN took it
5 positive, 36 neutral or asking, 3 mixed, 5 critical, out of 49 comments about the project. Thread score -0.01.
Execution: 0 for, 3 againstIdea: 2 for, 1 againstPerformance: 1 for, 0 againstDesign: 0 for, 1 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Useful for comparing OCR models; a long-awaited benchmark
From 3 comments
We've got like 10 LLM arenas but nothing for OCR yet, really hope this takes off!
ylhert on HNThis is a killer idea!
profburial on HNLove this! Would have liked to see something like textract for a pre-LLM benchmark (but of course that's expensive), and also a distinction between handwritten text and printed…
zzleeper on HN
Objections
Accuracy on complex tables is hard to judge; arenas are weak for correctness
From 2 comments
What is needed to evaluate OCR for most business applications (above everything else) is accuracy. Some results look plausible but are just plain wrong. That is worse than…
molf on HNThis is why arenas are generally a bad idea for assessing correctness in visual tasks.
fzysingularity on HN
Bias toward English; no per-language split
From 2 comments
I suggest you make explicit the assumption that this website is specifically about English text. Otherwise the leaderboard is pretty meaningless, with extreme differences in…
deaux on HNLMArena splits their leaderboard by language: maybe you should consider doing the same thing I assume to do that you’d need another model to do language detection on the inputs…
skissane on HN
Some models hang or loop, and the UI gets stuck
From 2 comments
A limitation of this leaderboard approach that I want to point out is that while the large general-purpose LLMs can make greater leaps of inference (on handwriting and poor…
daemonologist on HNReally like the idea. Unfortunately, my first upload is still spinning on one of the models about 5 minutes in. Clicking "Stop Battle" seems to do nothing either
ajmurmann on HN
Requests
Add more OCR models such as Mistral, Paddle, Chandra, Landing.ai DPT-2
From 5 comments
Most of these are general LLM’s and not specifically OCR models. Where is Google Vision, Mistral, Paddle, Nanonets, or Chandra??
codeddesign on HNPlease add Chandra by Datalab
ianhawes on HNAny plans to add Document Pre-trained transformer-2 (DPT-2) from https://landing.ai/?
densekernel on HNWe need to see Landing.ai DPT-2, from my tests its the best in term of ability to extract structure from complex tables so far.
coulix on HNWould be great to add MistralOCR!
timbmg on HN
Mobile UX fixes and MathJax rendering
From 2 comments
Two suggestions: UX on mobile isn’t great. It wasn’t obvious to me where the second model output was and I was thrown off even more so because the option to vote for model 1…
ComputerGuru on HNHey! Dev who made this here. I hear you on the mobile UX, it's on my docket of things to fix. Same with math plugin! Thanks for the suggestions.
rubikscubeguy on HN
Latency and cost metrics alongside quality
From 2 comments
FYI one of the models on the battle was pretty slow to load. Are these also being rated on latency or just quality?
fzysingularity on HNUltimately, there’s some intersection of accuracy x cost x speed that’s ideal, which can be different per use case. We’ll surface all of those metrics shortly so that you can…
kbyatnal on HN
Programmatic API access
From 1 comment
Is there a way I can invoke this programatically?
deeptishukla22 on HN
Questions
Will traditional OCR tools like Tesseract be added?
From 1 comment
There have been such a large number of OCR tools pop up over the past ~year; sorely in need for some benchmarks to compare them. Would love to see support for normal OCR tools…
cdrini on HN
Alternatives named
Chandra, MinerU, Azure Document Intelligence, Landing.ai DPT-2
From 4 comments
Please add Chandra by Datalab
ianhawes on HNcool UI and lets anyone upload a doc. but lacks https://github.com/opendatalab/mineru
vdm on HNNice! Would love to see Azure Document Intelligence on this
athoscouto on HNWe need to see Landing.ai DPT-2, from my tests its the best in term of ability to extract structure from complex tables so far.
coulix on HN