---
type: "post"
id: "only-18-mb-of-onnx-weights-no-pytorch-61535a"
slug: "only-18-mb-of-onnx-weights-no-pytorch-61535a"
title: "Only 18 MB of ONNX weights, no PyTorch; weights on Hugging Face ."
description: "Indra OCR makes local document extraction easier to deploy by packaging OCR weights and a usable interface in a small ONNX Runtime distribution."
retrieval_nugget: "Indra OCR makes local document extraction easier to deploy by packaging OCR weights and a usable interface in a small ONNX Runtime distribution."
published_at: "2026-10-17"
updated_at: "2026-10-17"
record_date: "2026-10-17"
date_kind: "published_at"
topics: ["local-models","models"]
source_urls: ["https://huggingface.co/datalab-to/indra-ocr"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:only-18-mb-of-onnx-weights-no-pytorch-61535a"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:1cdf630924107ee828cd60d57f37e8655adc88aac199c27f7e9e61b07cc649af","reviewed_at":"2026-10-17T14:03:04.949Z","source_evidence_count":1,"verified_claim_count":4,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":4,"implication_count":1,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"Indra OCR makes local document extraction easier to deploy by packaging OCR weights and a usable interface in a small ONNX Runtime distribution.","observed_facts":[{"text":"The model card describes about 18 MB of model files and says installation does not require PyTorch.","source_urls":["https://huggingface.co/datalab-to/indra-ocr"]},{"text":"The project provides CLI and Python entry points for PDFs and images, with Markdown, text, or JSON outputs that can include line boxes and labeled layout blocks.","source_urls":["https://huggingface.co/datalab-to/indra-ocr"]},{"text":"The model card reports up to 49 pages per second on CPU and a score of 41.3 on OmniParseBench Text.","source_urls":["https://huggingface.co/datalab-to/indra-ocr"]},{"text":"The card identifies Apache 2.0 for code and a modified OpenRAIL-M license for weights, with commercial-use restrictions described by user category.","source_urls":["https://huggingface.co/datalab-to/indra-ocr"]}],"mechanism":"The mechanism is a compact runtime package paired with outputs that preserve structure: Markdown, text, or JSON with line boxes and labeled layout blocks.","why_now":"Why now: a small CPU-oriented package gives document-processing teams another route to test OCR without bringing a PyTorch installation into the deployment.","implications":["The first implication is that teams can evaluate structured, locally processed OCR outputs with a lightweight setup rather than treating OCR as a hosted-only service."],"evidence_boundary":"The evidence boundary is the project’s model card: it documents interfaces, stated sizes, benchmark results, and licensing terms, but this evidence does not independently establish comparative performance or suitability for a given workload.","watch_conditions":["Watch whether independent tests on representative scans confirm the claimed speed and accuracy, and whether the weight-license terms fit intended commercial use."],"related_records":[],"new_branch_reason":"The release is distinct from earlier OCR records because this source describes a small local ONNX package and its deployment and licensing details, rather than a long-document parsing approach."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/only-18-mb-of-onnx-weights-no-pytorch-61535a/","markdown":"https://newruntime.com/posts/only-18-mb-of-onnx-weights-no-pytorch-61535a.md","json":"https://newruntime.com/posts/only-18-mb-of-onnx-weights-no-pytorch-61535a.json"}
related_shifts: ["document-processing-shifts-from-hand-crafted-ocr-pipelines-to-llm"]
---

# Only 18 MB of ONNX weights, no PyTorch; weights on Hugging Face .

## Retrieval answer

Indra OCR makes local document extraction easier to deploy by packaging OCR weights and a usable interface in a small ONNX Runtime distribution.

Indra OCR makes local document extraction easier to deploy by packaging OCR weights and a usable interface in a small ONNX Runtime distribution. That is a deployment proposition, not proof that it outperforms alternatives on every document corpus. The model card describes about 18 MB of model files, says installation does not require PyTorch, and provides CLI and Python entry points for PDF and image inputs.

The mechanism is a compact runtime package paired with outputs that preserve structure: Markdown, text, or JSON with line boxes and labeled layout blocks. Datalab reports up to 49 pages per second on CPU and a 41.3 score on OmniParseBench Text. Those figures make the release worth testing for teams that need a local document pipeline, but they remain vendor-reported claims in the supplied source. The card also separates Apache 2.0 code from weights under a modified OpenRAIL-M license, with commercial-use restrictions described by user category.

Why now: a small CPU-oriented package gives document-processing teams another route to test OCR without bringing a PyTorch installation into the deployment. The first implication is that teams can evaluate structured, locally processed OCR outputs with a lightweight setup rather than treating OCR as a hosted-only service. The evidence boundary is the project’s model card: it documents interfaces, stated sizes, benchmark results, and licensing terms, but this evidence does not independently establish comparative performance or suitability for a given workload. Watch whether independent tests on representative scans confirm the claimed speed and accuracy, and whether the weight-license terms fit intended commercial use. The release is distinct from earlier OCR records because this source describes a small local ONNX package and its deployment and licensing details, rather than a long-document parsing approach.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Local Models](https://newruntime.com/topics/local-models.md): Explore the local-models topic hub.
- topic hub: [Models](https://newruntime.com/topics/models.md): Explore the models topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)