One‑click Jupyter & VS Code
Boot into a ready-to-use environment with CUDA, PyTorch, and your SSH keys preloaded. Stop, snapshot, and resume without cold starts.
EU GPU Cloud
Built for AI Teams
Spin up powerful GPU instances in seconds. EU data residency by default, GDPR/AI‑Act aligned.
Research Labs
Universities
Startups
Enterprises
AI Teams
Data Science
mentronlabs – provision.sh
$ ml spinup --gpu H100 --region eu-central --jupyter
Creating VPC eu-central‑vpc‑01 … done
Provisioning node h100‑xlarge … done
Mounting volume projects:/mnt/data … done
Starting Jupyter at https://eu.mlnb.io/abcd
SSH: ssh ubuntu@eu.mlnb.io -p 2231
$ _
Features
Provision in seconds. Scale without drama.
A product-first experience that keeps the busywork out of your way so you can ship models faster—without compliance headaches.
EU‑hosted & GDPR/AI‑Act aligned
Choose Austria or Germany for data residency. Fine‑grained IAM, private networking, and audit logs as standard.
Predictable pricing
Simple hourly rates with optional committed discounts. Real‑time spend caps and alerts so your experiment never becomes a surprise bill.
Private networking by default
Project‑isolated VPCs, security groups, and inbound allow‑lists. Zero‑trust posture with just‑in‑time SSH.
Fast images for popular stacks
Curated images for PyTorch, JAX/Flax, TensorRT‑LLM, and vLLM. Bring your own container or bake a snapshot from any running node.
Team & enterprise ready
SSO/SAML, role‑based access, project budgets, and organization‑wide policies. Exportable usage reports for finance.
Pricing
Transparent hourly rates. Discounts for commitments.
Pick from on-demand or reserved. Stop any time and only pay for what you use. No hidden storage or egress surprises.
AMD RX 5700
€0.69/hr
Great for classical ML, preprocessing and light fine-tuning.
NVIDIA RTX 3060 Ti
€1.29/hr
Solid for LLM fine-tunes, diffusion, and mid-scale training.
Coming soon
NVIDIA A100 (40/80GB)
€—/hr
For production-grade training and high-throughput inference with NVLink.
Prices shown are examples; configure exact regions, storage, and networking at checkout.
FAQ
Pricing & usage — common questions
Short answers to help you decide quickly. Reach out if you need custom terms.
How is billing calculated?
Billing is per-second after a 1-minute minimum. You pay for GPU runtime plus any optional storage or egress you enable. Stop the instance and billing stops.
Do you offer reserved/committed discounts?
Yes. Commit to a monthly baseline (e.g., 100 GPU-hours) for a discount. We can tailor commitments by GPU family or region.
Spot vs. on-demand — what’s the difference?
On-demand is guaranteed capacity at the listed rate. Spot is cheaper but may be reclaimed; ideal for fault-tolerant training and batch inference.
Where is my data stored? EU residency?
Instances and volumes run in the EU (Austria/Germany). Data residency is EU by default; backups and logs remain within the EU regions you select.
What teams say
A fast path from idea to training run
Spin‑up to training in under five minutes. The cost alerts saved us on day one.
EU residency + SSO made security sign‑off painless. Our researchers love the Jupyter defaults.
Great 4090 availability for fine‑tuning. Snapshots help us keep experiments reproducible.
Simple pricing and the spending guardrails give our finance team peace of mind.
Security & Trust
Built with a zero‑trust posture
Private networking by default, audit logs, SSO/SAML, and encryption in transit & at rest. ISO 27001 roadmap, SOC 2 in progress.
SSO/SAML & RBAC
Bring your identity provider. Define roles per project and keep credentials time‑boxed.
Audit logs
Every action is recorded. Export logs for compliance or stream to your SIEM.