Hyperbots
A hosted, finance-specialised OCR service for document extraction.
Scores against the six-parser field
| Parser | Table |
|---|---|
| pdf-inspector | 100% |
| LiteParse | 100% |
| OpenDataLoader PDF | 0% |
| @opendocsg/pdf2md | 0% |
| unpdf | 0% |
| pdf-parse | 0% |
| Parser | Reading order |
|---|---|
| pdf-inspector | 100% |
| LiteParse | 100% |
| OpenDataLoader PDF | 76% |
| @opendocsg/pdf2md | 76% |
| unpdf | 76% |
| pdf-parse | 76% |
| Parser | Heading |
|---|---|
| pdf-inspector | 100% |
| LiteParse | 80% |
| OpenDataLoader PDF | 100% |
| @opendocsg/pdf2md | 100% |
| unpdf | 0% |
| pdf-parse | 0% |
Speed at scale against the field
| Parser | Median time |
|---|---|
| LiteParse | 54.0s |
| pdf-inspector | 2.9s |
| OpenDataLoader PDF | 2.7s |
| @opendocsg/pdf2md | 1.5s |
| pdf-parse | 0.6s |
| unpdf | 0.5s |
OCR routing signal
This parser exposes no page-level OCR signal — it does not flag which pages it believes need optical character recognition, so there is no precision, recall or reason breakdown to show. Only pdf-inspector and LiteParse publish that signal on this site.
No parser in this roster currently reads the scanned fixture (0 of 6) — recovering it would require optical character recognition applied to a full-page image, which none of these adapters do.
What this adapter measures
Evaluate a paid OCR path on financial documents, subject to the site's paid-run quota.
Not included in the free fixture benchmark. No scanned-document accuracy claim is published here.
Measured fixture results
How these checks workCaptured 2026-09-16 · darwin/arm64 · Node v22.21.1. Median of up to three sequential local samples. Timings will vary on other machines.
No published measurement for Hyperbots. No score or timing is inferred.