{"author":"adnan9999","children":[{"author":"adnan9999","children":[{"author":"warthog","children":[{"author":"adnan9999","children":[],"created_at":"2026-05-25T22:27:03.000Z","created_at_i":1779748023,"id":48272643,"options":[],"parent_id":48272607,"points":null,"story_id":48271937,"text":"Yeah, we&#x27;re not fully self-serve yet. Shoot me an email at adnan@unsiloed.ai with some info on the use case and I&#x27;ll get you set up on the platform.","title":null,"type":"comment","url":null}],"created_at":"2026-05-25T22:21:14.000Z","created_at_i":1779747674,"id":48272607,"options":[],"parent_id":48272382,"points":null,"story_id":48271937,"text":"website has no self serve sign up","title":null,"type":"comment","url":null}],"created_at":"2026-05-25T21:48:34.000Z","created_at_i":1779745714,"id":48272382,"options":[],"parent_id":48271937,"points":null,"story_id":48271937,"text":"Founder here. If you&#x27;ve got a notorious PDF you would like us to try , pls feel free to drop it in the comments. We&#x27;ll run it and share the output here.","title":null,"type":"comment","url":null},{"author":"pshishodia","children":[],"created_at":"2026-05-25T23:05:20.000Z","created_at_i":1779750320,"id":48272903,"options":[],"parent_id":48271937,"points":null,"story_id":48271937,"text":"Damn unreal that OCR was so unsolved until now","title":null,"type":"comment","url":null}],"created_at":"2026-05-25T21:35:03.000Z","created_at_i":1779744903,"id":48271937,"options":[],"parent_id":null,"points":9,"story_id":48271937,"text":"Most of the document parsers fail on real world challenges like complex tables, handwritten documents, historical document scans, equations, multi-column layouts, complex reading order, etc. We built Unsiloed Parser to handle exactly these cases.<p>Our latest parser v3.1 achieved #1 rank and scored 88.0 strict pass-rate on olmOCR-Bench. We ran the evaluation across 1,403 PDFs and 8,413 unit tests using the unmodified upstream Allen AI scorer (olmocr==0.4.27) and found Unsiloed beats 18 other OCR services, including GPT-5.5, Claude Opus 4.7, LlamaParse, Reducto, Azure Document Intelligence, AWS Textract, and Unstructured.<p>When we dug deeper into the failure cases, we found many errors were not OCR errors but things like \\frac vs \\dfrac, whitespace differences, or equivalent LaTeX renderings. We ran a secondary LLM-as-Judge evaluation to classify real misses vs semantic equivalents, which lifts the corrected score to 94.8 (explained deeply in the blog post).<p>Blog with full methodology and examples: <a href=\"https:&#x2F;&#x2F;www.unsiloed.ai&#x2F;blog&#x2F;unsiloed-ai-achieves-1-rank-on-olmocr-bench-2\">https:&#x2F;&#x2F;www.unsiloed.ai&#x2F;blog&#x2F;unsiloed-ai-achieves-1-rank-on-...</a><p>Evaluation Code for reproducibility:\n<a href=\"https:&#x2F;&#x2F;github.com&#x2F;Unsiloed-AI&#x2F;unsiloed-olmocr-benchmark\" rel=\"nofollow\">https:&#x2F;&#x2F;github.com&#x2F;Unsiloed-AI&#x2F;unsiloed-olmocr-benchmark</a><p>Feel free to post your messiest PDFs in the comment and we&#x27;ll run it through Unsiloed parser and share the output here.","title":"Show HN: Unsiloed AI \u2013 #1 on olmOCR-Bench","type":"story","url":null}
