A Docling alternative with bindings beyond Python.
This Docling alternative page compares licence, benchmark scores and community. We read competitor facts between 2 and 16 November 2026.
Competitor facts read between 2 and 16 November 2026.
PDFluent and Docling, row by row.
The benchmark's README lists docling at 0.882 overall against 0.9282 for PDFluent 1.4.0 quality auto. PDFluent is ahead on all 4 scores: overall, reading order, table structure and heading hierarchy. This is one benchmark and one run. The competitor figures come from its README.*
| Dimension | PDFluent | Docling |
|---|---|---|
| Licence | AGPL-3.0, or a commercial licence | MIT (Docling 2.150.0) |
| Languages | Rust, Python, Node.js, WebAssembly, Java and .NET, plus a command-line tool | A Python package on PyPI |
| opendataloader-bench, overallPDFluent ahead | 0.9282 (PDFluent 1.4.0, quality auto) | 0.882 |
| Reading orderPDFluent ahead | 0.9463 | 0.898 |
| Table structure (TEDS)PDFluent ahead | 0.9140 | 0.887 |
| Heading hierarchyPDFluent ahead | 0.8767 | 0.824 |
| Docling-shaped output | extract_docling returns Docling JSON. Structured extraction ships as a Preview. | Its own document format |
| Age and community (3 November 2026)Other tool ahead | pdfluent-sdk: 40 GitHub stars, first PyPI release 2026-05-14 | docling: 70,054 GitHub stars, first PyPI release 2024-07-15, created at IBM Research |
* Overall score 0.9282 (SDK 1.4.0, quality="auto", keep_furniture=True, layout, table and formula models installed) on all 200 documents, with the benchmark’s own harness and evaluator, measured 11 November 2026 on the final macOS arm64 release wheel. Competitor figures are taken from the benchmark’s public README as read on 3 November 2026 (highest overall there: 0.907, opendataloader [hybrid]), not re-run by us. The leaderboard engines ran on their own setups while PDFluent was measured on macOS arm64, and table structure depends on the platform. PDFluent is not on that leaderboard. Other engines have submissions open that are not yet on the leaderboard. One run, no confidence interval. The PDFluent row keeps page headers and footers because the benchmark’s ground truth counts them as content. PyMuPDF and MuPDF are trademarks of Artifex Software, Inc. PDFluent is an independent product of Innovation Trigger B.V. and is not affiliated with or endorsed by Artifex.
* GitHub stars and first PyPI release read on 3 November 2026. Stars measure attention, not quality, and a first release is not a founding date. IBM Research created Docling and donated it to the LF AI & Data Foundation on 29 April 2025.
Every engine, one scale.
- PDFluent 1.4.00.9282
- opendataloader [hybrid]0.907
- nutrient0.885
- docling0.882
- marker0.861
- unstructured [hi_res]0.841
- edgeparse0.837
- opendataloader0.831
- mineru0.831
- pymupdf4llm0.732
- unstructured0.686
- markitdown0.589
- liteparse0.576
PDFluent 1.4.0 with quality auto, page headers and footers kept. Other engines as published in the benchmark README on 2 November 2026. Footnote
The chart shows the 12 engines on the public opendataloader-bench leaderboard next to PDFluent 1.4.0 at quality auto. On the four scores of that benchmark PDFluent is ahead of Docling on 4 of 4. The table under the chart gives every engine on every score.
PDFluent and the 12 leaderboard engines on all four scores
| Engine | Overall | Reading order | Table structure | Headings |
|---|---|---|---|---|
| PDFluent 1.4.0 | 0.9282 | 0.9463 | 0.9140 | 0.8767 |
| opendataloader [hybrid] | 0.907 | 0.934 | 0.928 | 0.821 |
| nutrient | 0.885 | 0.925 | 0.708 | 0.819 |
| docling | 0.882 | 0.898 | 0.887 | 0.824 |
| marker | 0.861 | 0.890 | 0.808 | 0.796 |
| unstructured [hi_res] | 0.841 | 0.904 | 0.588 | 0.749 |
| edgeparse | 0.837 | 0.894 | 0.717 | 0.706 |
| opendataloader | 0.831 | 0.902 | 0.489 | 0.739 |
| mineru | 0.831 | 0.857 | 0.873 | 0.743 |
| pymupdf4llm | 0.732 | 0.885 | 0.401 | 0.412 |
| unstructured | 0.686 | 0.882 | 0.000 | 0.388 |
| markitdown | 0.589 | 0.844 | 0.273 | 0.000 |
| liteparse | 0.576 | 0.866 | 0.000 | 0.000 |
Their call, next to ours.
1from docling.document_converter import DocumentConverter23converter = DocumentConverter()4result = converter.convert("report.pdf")5print(result.document.export_to_markdown())
1from pdfluent import Document23# pip install pdfluent4with Document("report.pdf") as doc:5 markdown = doc.extract_markdown()6 docling_json = doc.extract_docling()
extract_docling takes first_page, page_count and name, and returns Docling JSON. quality="balanced" adds tables once the models are installed. The Docling code is the example from its own documentation, read on 16 November 2026.
What each costs, as of 16 November 2026.
Neither tool charges per page. Docling is MIT licensed. PDFluent is free under the AGPL-3.0, or a commercial licence costs EUR 999 per organisation per year excluding VAT as of 7 September 2026. The cost is the licence choice and your own compute.
| Product | Plan | Price |
|---|---|---|
| PDFluent SDK | Free under the AGPL-3.0, or a commercial licence | EUR 0, or EUR 999 per organisation per year, excluding VAT (as of 7 September 2026)No per-page fee. You run it on your own hardware. |
| PDFluent hosted API | Coming soon, not live yet | Planned: prepaid packs from USD 25 for 25,000 credits. 1 credit per page at fast, 3 at autoUSD 1.00 and USD 3.00 per 1,000 pages at the USD 25 pack. Planned terms as of 16 November 2026. |
| Docling | MIT licence | No price list: a Python package you run yourselfYour cost is your own compute, as with the PDFluent SDK. |
* List prices read on 16 November 2026 from each vendor’s own pricing page, and the arithmetic on them. The services name their tiers and strategies differently and have not been measured against each other, so a lower price is not a like for like saving. The PDFluent hosted API is coming soon and is not live; its prices are planned terms, checked again at launch. The SDK runs on your own hardware, which has its own cost. The SDK price is as of 7 September 2026.
Who each one suits.
Pick PDFluent when
PDFluent suits teams that work outside Python, with bindings for Rust, Node.js, WebAssembly, Java and .NET. The SDK runs inside your own process on your own machine. There is no cloud service, no licence server and no activation, and its default build has no HTTP client. Structured extraction returns Docling JSON, so Docling-shaped output stays available.
Pick Docling when
Docling is the better fit when you want an MIT-licensed tool from IBM Research. IBM Research created Docling and donated it to the Linux Foundation AI & Data Foundation on 29 April 2025. Docling also has the larger community: 70,054 GitHub stars on 3 November 2026, against 40 for pdfluent-sdk. Stars measure attention rather than quality.
Three questions.
Which licence do I need?+
Can I keep Docling-shaped output?+
How hard is it to switch from Docling?+
Try it on a
document you know.
Install it, open one file you can judge, and read what comes back.
Other comparisons: LlamaParse alternative · Unstructured alternative