Fully Managed Docling Serve
as a Service

Deploy Docling Serve as a fully managed service starting at €9/mo. Get automated backups, SSL, updates, support and monitoring included.

Docling Serve is an open-source HTTP API for document conversion. PDFs, Office files, images and audio become structured Markdown, HTML or JSON that a retrieval index or a language model can actually read. It combines the convenience of a hosted parsing API with the privacy and control of software you run yourself.

✓Free 7-day trial  ✓99.9% Uptime SLA  ✓No credit card  ✓Cancel anytime

✓Free 7-day trial  ✓99.9% Uptime SLA
✓No credit card  ✓Cancel anytime

Docling Serve

Docling Serve

STARTING AT

€9/month
Automated Backups
Monitoring
Automated Updates
Auto SSL

USAGE

Unlimited
Human Support
Custom Domains
Terminal Access
File Manager Access
Deploy in your region — 21 locations worldwide
GermanyFinlandNetherlandsUKSwedenUnited StatesCanadaSingaporeJapanAustraliaBrazilSouth Africa+9 more →
Docling Serve Preview Image

ABOUT THE SOFTWARE

What is Docling Serve

Docling Serve is the HTTP API around Docling, the open-source document conversion toolkit. It takes a file or a URL and returns structured output, Markdown, HTML, JSON or plain text, with layout, tables and reading order preserved rather than flattened into a wall of text.

Docling was started by the AI for knowledge team at IBM Research in Zurich and is now hosted as a project in the LF AI and Data Foundation, which means no single vendor controls its direction. Both the library and the server are released under the MIT licence, so nothing in the licence restricts who may run it or offer it to others.

The project releases frequently, with new versions landing every week or two, and the pieces are individually swappable. Table structure comes from a model called TableFormer, layout and reading order from a separate document model, and character recognition from whichever engine you select. A compact vision-language model, Granite-Docling, can do the whole conversion in a single pass instead.

FEATURES

What Docling Serve does

Docling Serve does one job, converting a document into structured output, through a small REST surface. What varies is the input format, the pipeline used to read it, and the shape of what comes back. These are the capabilities the software gives you.

Wide input coverage

PDF, DOCX, XLSX and PPTX and their legacy forms, open document formats, EPUB, HTML, Markdown, AsciiDoc, LaTeX, CSV, the usual raster image formats, email files, plus audio and video for transcription.

Structured output formats

Markdown, HTML with embedded or referenced images, lossless JSON holding the full document model, DocTags, plain text, WebVTT for audio, and pre-chunked output ready for a retrieval index.

Layout and reading order

A layout model assigns every element a type and a position, then produces reading order. Multi-column pages, captions, headers and footers come back labelled rather than interleaved at random.

Table structure recognition

TableFormer predicts rows, columns, spans and header hierarchy from a table crop, including borderless and partially ruled tables, then matches that prediction back to the text cells in the source file.

Character recognition

Scanned pages go through a pluggable engine, EasyOCR, Tesseract or RapidOCR, with per-request language selection and a choice between full-page recognition and reading only the regions that need it.

Vision-language pipeline

Instead of the modular stack, a single Granite-Docling pass reads the page end to end and emits DocTags, keeping tables, equations and code intact. It copes better with documents the modular pipeline finds ambiguous.

Chunking for retrieval

A hybrid chunker splits output into token-bounded pieces carrying page numbers and bounding boxes, so an answer assembled from a chunk can be traced back to the page it came from.

Sync and async REST endpoints

Single files go through a synchronous convert call. Batches go to an async endpoint that returns a task you poll. An optional browser interface sits alongside for manual conversions.

WHAT'S ALWAYS INCLUDED

Every app. Fully managed.
Nothing extra to pay for.

Every app on DANIAN ships with the same operational baseline. You do not choose a backup add-on or a monitoring tier. Hosting, patching, backups, SSL, and support are part of the €9, and they apply to Docling Serve the same way they apply to everything else in the catalog.

Automatic updates and patches

Apps run the latest stable version. Security patches applied silently, with rollback if needed.

Daily off-site backups

Multiple daily backups in redundant off-site locations. One-click restore if anything goes wrong.

24/7 uptime monitoring

Continuous monitoring with instant alerting. We respond before you notice.

SSL, firewall, DDoS protection

Auto-renewing SSL, hardened firewall rules, DDoS mitigation on every deployment.

Performance and scaling

We monitor resource usage continuously. When your app needs more headroom, we flag it and upgrade with your explicit approval.

Dedicated engineering support

Real engineers by email and ticket. DNS, SMTP & migration help. All included in €9.

WHY MANAGED

Why teams pick managed Docling Serve

When IBM published Granite-Docling under a permissive licence, running your own parser stopped being a quality compromise. A compact model that keeps tables, formulas and code intact removed the last argument for sending confidential documents to a per-page API. What remained was the running of it.

Running Docling Serve is not the same as installing it. The install is one container. The operation is a model cache that survives restarts, a worker configuration that matches the task engine, page and file ceilings that exist before somebody uploads a 900-page scan, and an API key, because the software ships without one.

The failure people hit first is the model download. Docling does not bake its weights into the base image; it fetches roughly 1.2 GB from a model registry on first use. A fresh container therefore either stalls on the first request, or, on a network with no outbound access, fails with a runtime error and no obvious cause. The upstream documentation is blunt about it: downloading models at startup is not recommended for production.

REVIEWS

Hear from customers ​like you​​​​​​​

Successful businesses and professionals around the world rely on DANIAN every day

USE CASES

Three teams who run Docling Serve on DANIAN

Docling Serve rarely runs alone. It sits between a pile of documents and something that needs them structured, a retrieval index, a bookkeeping system, an internal search box. These three shapes cover most of what people build with it.

5-PERSON AI PRODUCT TEAM

Feeding a retrieval chatbot documents where the answer lives in a table

Technical PDFs where the useful number is in a table, not a paragraph. They run the accurate table preset, chunk with page and bounding-box metadata so an answer can cite a page, and push results into a vector store. Region chosen next to their model host to keep round trips short.

OFFLINE RESEARCH GROUP

Converting confidential documents with no outbound access at all

Documents that are not permitted to reach a third-party API. Remote services stay off, model weights are pre-fetched to the artifacts path, and the instance talks to nothing outside their own network. They pull structured JSON on a schedule and index it locally.

6-PERSON ACCOUNTING PRACTICE

Turning a month of client invoices into line items nobody retypes

Scanned invoices and receipts of uneven quality. Full-page recognition is on because the scans are inconsistent, the file-size ceiling is raised for batch uploads, and the async endpoint takes the month-end run while the sync endpoint handles single documents during the day.

COMPARISON

Four ways to run Docling Serve

Document parsing is billed by the page almost everywhere, so the cost of a hosted API tracks how much work you do rather than how many people do it. This compares four ways to get the same structured output, priced at three monthly page volumes.

Path1,000 pages5,000 pages10,000 pagesWhat you also take on
Azure AI Document Intelligence, layout model$10$50$100Billed per page, so the bill grows with the work. Documents leave your network to be processed.
Self-host on a $24/mo production-class VPS$24$24$24Your hours: model cache, worker tuning, ceilings, certificate renewal, upgrades, and being the one who gets paged.
Home server, refurbished tower class~$500 once+ ~$9/mo powersame hardwareHardware failure is yours to fix. Residential upload speed caps how fast a batch finishes.
DANIAN managed Docling Serve9 euro9 euro9 euroBackups, updates, monitoring, TLS and support included. You pick the region.

The crossover is not subtle. At a thousand pages a month the hosted API is close enough that convenience wins the argument. At ten thousand it is roughly eleven times the flat price, every month, and the gap widens with every page you add.

Cost is only half of it. The hosted API has your documents processed by someone else, which is a conversation you may not want to have with a client. The other two keep the documents in-house but hand you the model cache, the ceilings, the renewals and the pager. Only one path does both.

BY INDUSTRY

Docling Serve for specific industries

Document parsing inherits the rules that govern the documents. A parser sitting in front of case files, tax records, client statements has to answer for where those files went and how long the output is kept. These three sectors put the sharpest demands on it.

Federal Rule of Civil Procedure 26 governs the discovery of electronically stored information, and Rule 37(e) provides the safe harbour for information lost through the routine, good-faith operation of a system, which makes the deletion behaviour of every tool in the chain a question somebody eventually asks. For a litigation support team we keep single-use result retrieval on with its purge delay, so converted output does not linger in a scratch directory once it has been collected, and we keep remote services off so privileged material never leaves the instance. A typical run converts a production set of scanned PDFs into JSON carrying page numbers and bounding boxes, so a passage quoted in a brief points back to a specific page of a specific document. A production of twenty thousand pages costs the same nine euros as a single page.

The retention arithmetic here is unusually specific. Records are generally kept three years, six where gross income is understated by more than a quarter, seven for worthless-securities and bad-debt claims, and four for employment-tax records, so a practice holds several overlapping horizons on the same client at once. For an accounting practice we raise the synchronous wait ceiling and set the file-size limit high enough for a batch upload, because month-end arrives as one large drop rather than a steady trickle. The workflow is invoices and receipts scanned at uneven quality, pushed through full-page recognition, and returned as JSON line items a bookkeeping system reads without retyping. That seven-year horizon is what makes structured output worth more than a stored image.

SEC Rule 17a-4 requires specified broker-dealer records to be preserved for six years, the first two of them in an easily accessible place, a two-speed obligation that only works if the records are searchable rather than merely stored. For a back office we set page and file ceilings so a bulk statement run cannot occupy the worker queue, and we keep the API key requirement on because the endpoint is called by an internal system rather than a browser. The workflow is monthly custodian statements and account-opening paperwork converted into structured JSON that an internal index can query field by field. Two years of accessible records for a mid-size book runs to hundreds of thousands of pages, and the price does not move.

FAQ

Frequently asked questions

Three groups of questions: how the software behaves and what we configure on your instance, how you move onto it from something else, and how billing, support and the platform work. All of it is on the page rather than behind a click.

01

Technical and configuration

Yes, and we set it up that way. We pre-fetch the model artifacts into a persistent path and point the artifacts path variable at it, so the instance does not depend on a third-party registry being reachable when it restarts.
Tables go through TableFormer, which predicts rows, columns, spans and header hierarchy from the table image and matches that back to the text cells. Borderless and partially ruled tables are handled. Pseudo-tables drawn with spaces stay hard.
EasyOCR, Tesseract and RapidOCR are available and selected per request. Between them the language coverage runs past a hundred, and the language is set on the request rather than at install time, so a mixed-language corpus does not need separate instances.
The standard pipeline is modular, with a layout model, a table model and a recognition engine each doing one job. The vision-language pipeline sends the page to Granite-Docling in one pass. Modular is faster on clean files; the single pass copes better with awkward layouts.

02

Migration and onboarding

21 datacenter locations across six continents. You choose the region at provisioning. Application data sits in the region you choose; pick whichever is closest to your users or matches your data-residency preference.
Yes. Request a region migration from the dashboard and we run the move in the background. The system emails you when the migration completes; total transfer time depends on data volume but typical instances finish in a few hours. There is no extra charge for a region change.
Yes. Full data export is available at any time, in a portable format you can bring to any infrastructure.
Point your client at your instance and send the API key in the header. The convert request takes a file or a URL and returns the format you ask for, so the change is usually the base URL, the auth header, and a re-read of the response shape.
Not character for character, because no two parsers agree. Docling returns Markdown, HTML, plain text or its own JSON document model. Convert a representative sample during the trial and compare before you rewire anything downstream.
That is what the trial is for. Run your worst documents through it, the scanned ones, the multi-column ones, the ones with the tables that break everything else, rather than the clean sample that makes every parser look good.
Yes. Point a record at your instance and we issue and renew the certificate. Custom domains are included rather than an upgrade. If you would rather keep it on a subdomain we provide, that works too.
Docling has documented integrations with the common retrieval frameworks and chat front-ends, and the API is plain HTTP either way. Most people call the convert endpoint from their own ingestion job rather than using a framework adapter at all.

03

Billing, support, and platform

€9 covers everything we do for that app: hardware in the region you choose, daily off-site backups with one-click restore, automatic security patches and version upgrades, 24/7 monitoring, SSL and firewall, and engineering support on email and the ticket section in the client portal. There are no setup fees or hidden line items. For more info see our Pricing page.
If you decide to continue, we charge €9/app/month from day 8. If you don't, the trial ends and you can export your data. No card is required for the trial, and we never auto-charge you without explicit consent.
No. The €9/month is flat regardless of how many users log into your app. Add 5 users or 50; the price doesn't change.
24/7 email and ticket support, both staffed by engineers who run the systems. We handle DNS configuration, SMTP setup, app integrations, performance tuning, troubleshooting, and migration help. Response time is typically under an hour. There is no tier system — every customer gets the same support.
Yes. Cancel from the dashboard. We don't charge a cancellation fee, we don't lock data, and we will export your data to you on request before deletion. data to you on request before deletion.
Every customer instance is backed up daily to a separate region from the primary. We test restores. You can request a restore at any backup point within the retention window — usually 7 days for daily backups.
Your application data sits in the region you choose at provisioning — 21 datacenter locations across six continents. Account-level data (billing, account email, support ticket history) is processed centrally. Application data region is picked by you, per app.
99.9% uptime SLA on every app, every tenant. Service credits are documented at danian.co/service-level-agreement. The status page is located at status.danian.co.
When your tenant approaches the resource ceiling — the base tier holds 1 vCPU/RAM, 30 GB storage — we notify you. Resource upgrades happen with your explicit consent; we will not upgrade your tenant or charge you without it.
We wait. We don't suspend the app or delete your data on the first failed charge. We email you, you fix the card on file, and we continue.
Invoices can be downloaded from the billing dashboard in PDF the day each charge succeeds. EU VAT is added where applicable and the VAT-reverse-charge regime applies for VAT-registered businesses with a valid number.
150+ open-source apps across automation, team chat, file sync, analytics, AI, password management, email marketing, dev tools, project management, smart home, CMS, and federated social. See the full catalog →
Yes. Every instance comes with a web-based terminal and a file manager in your DANIAN management dashboard. Useful for managing your data and customizations.
Resources scale with your usage. If your app needs more vCPU, RAM, or storage, we add it — and we ask first before any change to your plan. €9 is the floor; resource-heavy workloads may price higher, but you'll always know in advance.
Yes. We have both a Partner program and an Affiliate program available. Anybody can sign up.
No contract. No minimum commitment. Cancel anytime from the dashboard with one click. The 7-day free trial requires no credit card. After the trial converts to paid, you can still cancel at any month without notice or penalty.

DEPLOY IN YOUR REGION

21 datacenter locations on six continents

Pick the region closest to your users.

United States, Germany, Finland, Singapore, Australia, Brazil, Canada, Netherlands, UK, Spain, Italy, France, Sweden, Malaysia, India, Japan, Mexico, Poland, South Korea, Chile, South Africa and more coming soon

Global Reach Map

Try managed Docling Serve for 7 days

No card. Cancel from the dashboard.