Sign inGet started

Find the best AI for a job

Describe what you want to get done in your own words. We rank the best-fitting models, datasets, papers and open-source tools in the Silytics index, and say why each one is on the list.

Understood as: summarize, long, documents, summarization, summary, document, ocr, pdf · task: summarization

Best models

Ranked by fit to the task, adoption and reported results. Hosted models are paid APIs; the rest can be downloaded.

  1. 1
    pszemraj/long-t5-tglobal-base-sci-simplify

    matches long, summarization · built for summarization · 221,149 downloads · apache-2.0 license

  2. 2
    knkarthick/MEETING_SUMMARY

    matches summarization, summary · built for summarization · 61,393 downloads · 40 reported benchmark results · apache-2.0 license

  3. 3
    cnicu/t5-small-booksum

    matches summarization, summary · built for summarization · 17,778 downloads · mit license

  4. 4
    facebook/bart-large-cnn

    matches summarization · built for summarization · 1,071,198 downloads · 6 reported benchmark results · 4 entries link to it · mit license

  5. 5
    sshleifer/distilbart-cnn-12-6

    matches summarization · built for summarization · 316,925 downloads · 3 entries link to it · apache-2.0 license

  6. 6
    lightonai/LightOnOCR-2-1B

    matches document, ocr, pdf · 876,278 downloads · 14 entries link to it · apache-2.0 license

Research to read

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6

Open-source tools and code

  1. 1
    officeparser

    matches document, ocr, pdf · 4,571,401 GitHub stars · MIT license

  2. 2
    PyMuPDF

    matches documents, document, ocr, pdf · 11,076 GitHub stars · AGPL-3.0 license

  3. 3
    papermerge

    matches documents, document, ocr, pdf · 2,934 GitHub stars · Apache-2.0 license

  4. 4
    open-paperless

    matches documents, document, ocr, pdf · 2,559 GitHub stars

  5. 5
    kordoc

    matches document, ocr, pdf · 221,862 GitHub stars · MIT license

  6. 6
    @json-render/react-pdf

    matches documents, document, pdf · 168,425 GitHub stars · Apache-2.0 license

Datasets

  1. 1
    AmazonScience/document-haystack

    matches long, document, pdf · 117,279 downloads

  2. 2
    llamaindex/ParseBench

    matches document, ocr, pdf · 47,408 downloads · apache-2.0 license

  3. 3
    gutoportelaa/dom-pi-pdfs-2025

    matches document, ocr, pdf · 42,320 downloads · cc-by-4.0 license

  4. 4
    Forceless/Zenodo10K

    matches document, pdf · 167,146 downloads

  5. 5
    mercor/apex-agents

    matches long, document · 171,048 downloads · cc-by-4.0 license

  6. 6
    allenai/olmOCR-bench

    matches document, ocr · 40,680 downloads · odc-by license

Rankings are computed from public metadata and are a starting point, not a guarantee. Always test on your own data and check each license. Ask the community if you are unsure.