# Classify documents with open-weight VLMs through a TypeSafe-compatible API — retained record

> Retained reddit record from reddit:computervision.

Canonical: https://highsignal.app/data/records/b45bb6a65e8bf9d6

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:computervision)  published 2026-09-29 16:54 ingested 2026-09-30 02:38

[original source ↗](https://www.reddit.com/r/computervision/comments/1wtfh37/classify_documents_with_openweight_vlms_through_a/)

## retained text



```
Ran the full 1,651-page OmniDocBench through the [VLM Run gateway](https://vlm.run/gateway) for 25 cents ($1 covers 6700+ pages). The API reads the page as pixels (no OCR->LLM hacks) and returns one typed, calibrated decision per page, \~180ms p50 e2e.

  
Curious what others are finding in terms of TypeSafe-compatible APIs for vision. Let me know what you think!
```



document fetched 2026-09-30 00:24 · canonical https://reddit.com/r/computervision/comments/1wtfh37/classify_documents_with_openweight_vlms_through_a
