Nicheloom

Market intelligence for builders — see what's gaining traction before it's crowded.

Unsiloed AI

#1 on olmOCR-Bench

Details

External ID
48271937
Source
HN
Company
—
Product
—
Website domain
—
Launched
May 25, 2026
Cohort
—
Upvotes
9
Upvotes percentile
0.5516962843295639
Tags
—
Fetched at
Sept. 7, 2026, 9:26 p.m.
Updated at
Sept. 7, 2026, 9:26 p.m.

Description

Most of the document parsers fail on real world challenges like complex tables, handwritten documents, historical document scans, equations, multi-column layouts, complex reading order, etc.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.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.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).Blog with full methodology and examples: https://www.unsiloed.ai/blog/unsiloed-ai-achieves-1-rank-on-...Evaluation Code for reproducibility: https://github.com/Unsiloed-AI/unsiloed-olmocr-benchmarkFeel free to post your messiest PDFs in the comment and we'll run it through Unsiloed parser and share the output here.

Enrichment

Theme
lightweight and on-device AI runtimes
Vertical
Horizontal
Function
Dev tools
Audience
Developer
AI stance
AI feature
Project type
Commercial product
Normalized one-liner
ocr benchmarking tool
Manually corrected
False

Could you build this?

No Achieving state-of-the-art benchmark results on complex document layout OCR requires proprietary model architectures, massive multimodal pre-training, and extensive synthetic/curated datasets.

What it would actually take: A real version requires training a specialized vision-language model (VLM) or custom layout transformer on millions of high-resolution pages containing complex tables, math equations, and non-standard reading orders. The pipeline involves custom synthetic data generation (rendering LaTeX, complex HTML tables), layout segmentation models, and high-performance inference servers (TensorRT-LLM / vLLM) on high-end GPU clusters. This requires deep machine learning research expertise and substantial compute budgets.

Discussion

4 comments analyzed.

Concerns raised: No self-serve signup available, Manual setup required via email

Competitors

Other products that read as similar to this one — 91 launches clear the similarity bar, closest 8 shown.

Attention rank: #36 of 92 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).

Launched 206 days after the earliest competitor.

Other launches for this product

Same idea, different domain

Nobody's really built a dev tools tool for Sales yet.