About LeemerLabs
Inference infrastructure,
built in Ireland.
LeemerLabs is the inference company of the Leemer Group. We run frontier open-weight models on Nvidia H200s in Ireland, behind an OpenAI-compatible gateway, with a free tier and a data path that never leaves the EU.
Founded
Waterford, Ireland
Parent
Leemer Group
Hardware
Nvidia H200 cluster
Data region
Dublin + Waterford, EU
Pricing
Free tier, paid for scale
API surface
OpenAI-compatible gateway
Why we exist
Our own products outgrew generic APIs.
LeemerLabs exists because Foundry exists. Training custom models for our own products and for clients meant we needed serious, European inference — not a US API with EU marketing. So we built it.
The infrastructure that serves our custom models is the same infrastructure that now serves a public gateway. Free tier for anyone building in Europe who wants a local, GDPR-native path to frontier models.
We are small on purpose. One small team, one clear remit: European inference that is legible, dependable, and priced like infrastructure — not like a rental.
What LeemerLabs is
Four primitives.
European by construction
Every request is served from EU infrastructure. Data residency isn't a compliance checkbox — it's how the gateway is built. No silent trans-Atlantic routing.
GDPR-native
Purpose limitation, deletion rights, data minimisation and sovereignty are primitives in the system, not policy documents stapled to a US API.
Frontier hardware
Nvidia H200s with 141GB of HBM3e. Enough memory to serve modern MoE models without weight streaming latency. Hosted locally.
OpenAI-compatible
A clean API surface that slots into the tools you already use. Point your client at api.leemerlabs.ie/v1 and keep shipping.
Frontier hardware
Nvidia H200
141 GB HBM3e · 4.8 TB/s
We chose H200s because they're the right tool for modern mixture-of-experts inference. Enough HBM3e memory to serve 235B-parameter MoE models without weight streaming penalties. Hosted locally so European traffic never leaves European compute.
Get in touch
A Waterford company, one email away.
Whether you want API access, a custom model, or just to talk about European inference — we reply within one working day.