When the document is the sensitive part
For plenty of teams the blocker was never rendering quality; it was sending salary, balance or diagnosis data to somebody else's server at all. Self-hosted KimiDoc removes the question: there is no third party.
Data residency you can point at
Run it in the region, datacentre or rack you are already accountable for. No cross-border transfer to document, because there is no transfer.
Nothing phones home
No telemetry, no licence check-in, no usage beacon. Remote fetching is off by default and the service runs air-gapped where required.
Auditable by inspection
One service, one database file, one storage directory. Your security team can read the config and watch the ports instead of trusting a questionnaire.
Unlimited by design
No document metering, no per-seat pricing. Retention rules are yours to set; capacity is whatever the machine you chose can do.
The identical engine, not a lesser edition
Everything the hosted API does ships in the install: the template editor with live preview, 100+ starter templates, batch jobs to 40,000 documents with retry and manifests, AES-256 encryption, watermarking, webhooks, multi-tenant accounts if you serve customers of your own, and AI template generation against any OpenAI-compatible endpoint you configure, including a local model. A fixed pre-warmed render pool keeps one modest server fast: requests never pay a browser start-up.
Install is one package
KimiDoc ships as a snap: dependencies contained, strictly confined, automatic security updates from the base system, one command to install and one file of configuration. Your licence email carries the exact steps; a first render is typically minutes after the download.
Questions teams ask
Is there a free self-hosted version?
No. Self-hosted KimiDoc is licensed per server, per year. The hosted API has a free tier (100 documents a month) that is the quickest way to evaluate the engine before an install, since both run identical rendering.
What are the system requirements?
One Linux server. The service ships as a snap with its dependencies contained, runs fully confined, and a modest virtual machine handles serious volume because rendering comes from a fixed pre-warmed pool rather than per-document browser processes.
Does it phone home or need internet access?
No telemetry, no licence check-in, no usage beacons. Remote fetching by templates is off by default, and the service runs air-gapped; updates are applied when you choose.
How is it different from open-source converters?
Projects like Gotenberg or Puppeteer-based containers give you a free rendering endpoint, and for many teams that is enough. KimiDoc adds the product around the engine: stored templates with sample data, a live editor, batch jobs with retry and manifests, encryption and watermarking, per-key metering, and multi-tenant accounts, plus a vendor accountable for all of it.
Can we run it multi-tenant for our own customers?
Yes. The same accounts mode that runs kimidoc.com ships in the product: signup, per-tenant isolation, API keys and quotas, on your hardware.
Tell us about your setup
Licensing is per server, per year. Write with your expected volume and environment and we will come back with the build, the steps and a price.