Unpublished collections stay unpublished.
Run AI on a machine you control.
Private LLM hosting on a dedicated machine in EU-Central, for museums, archives, and cultural institutions cataloguing, transcribing, and researching material before it's ready for public release.
Not legal or archival-practice advice. This page describes GPUwerk's infrastructure and contractual commitments. Copyright status, donor restrictions, and what your institution may do with unpublished material are questions for your registrar, legal counsel, or governing body.
Why museum and archive work doesn't fit a public AI tool
Collections work routinely touches material that isn't cleared for public view: unaccessioned donations, provenance research still in progress, correspondence under a donor's confidentiality terms, and loan agreements naming private collectors. Pasting any of that into a consumer chat tool sends it to a third party with no contract governing what happens next.
"We just need help transcribing handwriting"
Transcription and OCR cleanup for correspondence, ledgers, and field notes is one of the more common uses of LLMs in archives work. If the material is unpublished or under embargo, running that transcription on a public API means the text has left the building before the institution has decided whether it should.
"Donor records are just names and addresses"
Donor and lender files often carry more than contact details: valuation figures, family history, sometimes health or financial context volunteered in correspondence. A dedicated instance keeps that material inside your access-controlled environment instead of a shared cloud tenancy.
"Our researchers already use public AI tools informally"
Curators and researchers drafting catalogue entries or summarizing acquisition files in a public tool is a real exposure, not a hypothetical one. A sanctioned alternative on your own instance, same chat interface, keeps that workflow off third-party infrastructure. See the private ChatGPT setup →
What runs well on a Spark node
Cataloguing assistance, transcription cleanup, provenance research summarization, and drafting exhibition or grant text from internal notes are typical workloads. Nothing here requires a data center GPU cluster; a single dedicated node handles most institutional collection sizes.
| Question | Answer |
|---|---|
| Where is data physically processed? | EU-Central, on a dedicated single-tenant machine assigned to your institution |
| Who can access instance content? | Through the instance itself, only holders of your SSH keys; password login is disabled fleet-wide. GPUwerk keeps infrastructure administrator access to the machine, as on any hosted service, and under the DPA does not use it on your content except at your request for support or where a legal obligation requires it. |
| Sub-processors for the workload? | None, listed at /legal/sub-processors |
| DPA (GDPR Art. 28)? | Published at /legal/dpa, no charge |
| Certifications held? | None claimed on this page or elsewhere on the site. If a funding body or governing trust requires a specific certification, verify directly with us before relying on this page. |
| On-demand rate | $0.79/hour (spark-1x), $1.79/hour (spark-2x). See /pricing for held/reserved rates. |
Questions we get from museums and archives
Can we use this to work with unpublished archive material?
Yes. The instance is a dedicated single-tenant machine assigned to your institution, so processing unpublished manuscripts, correspondence, or object files with an LLM stays on hardware you control rather than a shared public service.
Does GPUwerk claim any rights over the material we process?
No. GPUwerk runs infrastructure, not the model or the content on it. Under the data processing agreement, GPUwerk does not access instance content except at your request for support or where a legal obligation requires it. Copyright and licensing questions about your collections are a matter for your institution and rights holders, not for us.
Is donor and loan agreement data covered by GDPR here?
Personal data about donors, lenders, or living artists is ordinary personal data under GDPR in most cases, and the same Article 28 DPA and EU-Central location apply. Whether any of it counts as special-category data depends on the specific content, which is for your institution to assess.
Do you sign a DPA for institutional records?
Yes, a standard GDPR Article 28 DPA is published at /legal/dpa at no charge, and there are no sub-processors for instance workloads.
Related pages
See how other institutions and creative businesses use private LLM hosting: event and conference organizers, print and prepress companies, and translation and localization agencies. Or start from the private LLM hosting overview.
Related pages
Bring your collections work to a machine you control.
Talk to the people who run the racks, or start with a pilot and a practical rollout plan.
Talk to us Deploy an instance