Docling Serve
Docling document conversion (PDF, Office, images to Markdown/JSON) as a self-hosted API with a demo UI.
Quick answer
Docling Serve is a docling document conversion (PDF, Office, images to Markdown/JSON) as a self-hosted API with a demo UI, listed under Document processing. A typical small-business use: a shared internal API that turns contracts, invoices and reports into Markdown for RAG pipelines. The project publishes no official RAM figure, so we do not estimate a monthly server cost. License: MIT, which allows internal business use.
Last updated · facts verified against official docs on · sources
Activity data comes from GitHub and was last refreshed Oct 28, 2026.
What is Docling Serve?
Docling Serve runs IBM's Docling toolkit as a FastAPI service. It converts PDFs, Office files, HTML and images into Markdown, JSON or the DoclingDocument format, with layout analysis, table structure and OCR, through a REST API. It ships container images for CPU and CUDA (a ROCm image can be built locally) and an optional /ui playground.
Where it fits in a 10–200 person business
- A shared internal API that turns contracts, invoices and reports into Markdown for RAG pipelines
- Batch conversion of scanned PDFs with OCR and table extraction without sending files to a cloud API
- Document parsing step behind n8n, Dify or custom scripts
These use cases are our editorial suggestions, not claims made by the project.
Can a business use Docling Serve? (license)
| License | MIT |
|---|---|
| Type | Permissive: MIT / Apache-2.0: use, modify and sell with few obligations. |
| Internal business use | Allowed under the license text |
| What to watch | Permissive. Internal and commercial use allowed; keep the license notice. |
This is a plain-English summary of the license text, not legal advice. Third-party models you run with it have their own licenses.
Data & compliance: does Docling Serve phone home?
| Telemetry (usage data sent to the vendor) | Not documented The documented telemetry options are OpenTelemetry metrics, traces and a Prometheus /metrics endpoint for your own monitoring (metrics on by default, traces off). Nothing in the docs describes data sent to the Docling project. Source: github.com/docling-project/docling-serve/blob/main/docs/configuration.md |
|---|---|
| Other calls home by default (updates, marketplaces, default cloud services) | Not documented Pipeline components may not make remote connections unless DOCLING_SERVE_ENABLE_REMOTE_SERVICES is enabled (default false), for example for API-hosted vision models. Source: github.com/docling-project/docling-serve/blob/main/docs/configuration.md |
| Single sign-on (OIDC / SAML / LDAP) | Not documented |
| Audit log | Not documented |
| Role-based access control (RBAC) | Not documented |
| Data residency / GDPR notes from the project | Not documented |
Checked against the project's own docs and repository on . "Not documented" means we could not find the answer in official sources; it does not mean the feature is missing or that no data is sent. Third-party services you connect (model APIs, vector databases, search engines) have their own data policies. Compare all apps.
What hardware does Docling Serve need?
The project does not publish a minimum CPU/RAM figure, so we leave the official value blank rather than guess.
GPU: CPU-only image (docling-serve-cpu) is published; CUDA 12.8 and 13.0 images are available for NVIDIA GPUs, and an AMD ROCm image can be built locally. DOCLING_DEVICE picks cpu, cuda or mps, otherwise the best device is chosen.
No CPU, RAM or disk minimums are published. Published image sizes: 4.4 GB (CPU-only and arm64 base), 8.7 GB (amd64 base), 11.4 GB (CUDA 12.8).
Source: github.com/docling-project/docling-serve#container-images
How much does it cost to host Docling Serve per month?
The project publishes no RAM figure we could verify, so we do not estimate a cost.
How to deploy Docling Serve
Method: Container (Podman or Docker)
podman run -p 5001:5001 -e DOCLING_SERVE_ENABLE_UI=1 quay.io/docling-project/docling-serve
# API on :5001, API docs on /docs, UI playground on /ui
# CPU-only variant: quay.io/docling-project/docling-serve-cpuFrom the official instructions: github.com/docling-project/docling-serve#getting-started. Always check the source for the current version before you run it.
Known gotchas when self-hosting Docling Serve
- No API key is set by default. Set DOCLING_SERVE_API_KEY so every request must send the X-Api-Key header; the API docs pages (/docs, /openapi.json) stay open unless DOCLING_SERVE_ENABLE_API_DOCS=false. source
- Results are single-use by default (DOCLING_SERVE_SINGLE_USE_RESULTS=true): an async result can be fetched only once. Setting it to false makes results pile up in the scratch directory. source
- CUDA images have no latest tag; pin an explicit version such as v1.18.0. The ROCm image is not published and must be built from source. source
- Models are loaded from DOCLING_SERVE_ARTIFACTS_PATH; a missing model raises a runtime error. Use persistent storage and pre-download models for production and multi-worker setups. source
Alternatives in Document processing
- PaddleOCR OCR toolkit and document AI engine that turns scans, images and PDFs into LLM-ready text, Markdown or JSON. Compare Docling Serve vs PaddleOCR
- MinerU Document parser that converts PDFs, scans and Office files into clean Markdown or JSON for LLMs and RAG. Compare Docling Serve vs MinerU
- Unstract No-code LLM platform that turns invoices, forms and other documents into structured JSON. Compare Docling Serve vs Unstract
- Paperless-AI AI auto-tagging and document chat for Paperless-ngx (currently unmaintained). Compare Docling Serve vs Paperless-AI
- paperless-gpt LLM add-on for paperless-ngx that does better OCR and writes titles, tags and correspondents. Compare Docling Serve vs paperless-gpt
Docling Serve FAQ
How much RAM does Docling Serve need?
Docling Serve does not publish an official RAM figure, so we do not state one. No CPU, RAM or disk minimums are published. Published image sizes: 4.4 GB (CPU-only and arm64 base), 8.7 GB (amd64 base), 11.4 GB (CUDA 12.8).
Does Docling Serve need a GPU?
CPU-only image (docling-serve-cpu) is published; CUDA 12.8 and 13.0 images are available for NVIDIA GPUs, and an AMD ROCm image can be built locally. DOCLING_DEVICE picks cpu, cuda or mps, otherwise the best device is chosen.
How much does it cost to self-host Docling Serve?
The software is free to download. The project publishes no official RAM figure, so we do not estimate a monthly server cost.
Is Docling Serve free for commercial use?
Docling Serve is licensed under MIT (Permissive). Internal business use is allowed under the license text. Permissive. Internal and commercial use allowed; keep the license notice. This is a summary, not legal advice.
What are common problems when self-hosting Docling Serve?
No API key is set by default. Set DOCLING_SERVE_API_KEY so every request must send the X-Api-Key header; the API docs pages (/docs, /openapi.json) stay open unless DOCLING_SERVE_ENABLE_API_DOCS=false. Results are single-use by default (DOCLING_SERVE_SINGLE_USE_RESULTS=true): an async result can be fetched only once. Setting it to false makes results pile up in the scratch directory. CUDA images have no latest tag; pin an explicit version such as v1.18.0. The ROCm image is not published and must be built from source. Models are loaded from DOCLING_SERVE_ARTIFACTS_PATH; a missing model raises a runtime error. Use persistent storage and pre-download models for production and multi-worker setups.
Is Docling Serve still maintained?
Yes, it is actively developed. Last commit: Oct 27, 2026. Latest release: v1.38.2 (Oct 27, 2026). Checked Oct 28, 2026.
How do I install Docling Serve?
The official method is Container (Podman or Docker). The exact commands are in the "How to deploy" section above and in the official instructions: https://github.com/docling-project/docling-serve#getting-started
What are self-hosted alternatives to Docling Serve?
Other apps in Document processing: PaddleOCR, MinerU, Unstract, Paperless-AI, paperless-gpt.
Sources
- GitHub repository: github.com/docling-project/docling-serve
- Official documentation: github.com/docling-project/docling-serve/blob/main/docs/README.md
- License text: github.com/docling-project/docling-serve/blob/main/LICENSE
- Hardware requirements, Gotcha 3: github.com/docling-project/docling-serve#container-images
- Deploy instructions: github.com/docling-project/docling-serve#getting-started
- Gotcha 1, Gotcha 2: github.com/docling-project/docling-serve/blob/main/docs/configuration.md
- Gotcha 4: github.com/docling-project/docling-serve/blob/main/docs/models.md
- Hetzner Cloud pricing (cost estimate): www.hetzner.com/cloud/regular-performance
- GitHub stars, commits and releases: GitHub REST API, fetched
Facts on this page were checked against the project's own repository and docs on 2026-10-28. Spotted something out of date? Tell us.