Contabo has launched its new GPU VPS platform, introducing dedicated NVIDIA GPU resources to its VPS portfolio for workloads requiring accelerated computing. Each GPU VPS instance includes a dedicated NVIDIA RTX PRO 6000 GPU based on NVIDIA’s Blackwell architecture, with 96 GB of GDDR7 VRAM and 24,064 CUDA cores.
The GPU features fifth-generation Tensor Cores with FP4 support, designed to accelerate AI and machine learning workloads while allowing larger models and longer-context applications to run on a single GPU. According to Contabo, the 96 GB of VRAM can accommodate models with up to 70 billion parameters on a single card, depending on the model architecture and workload.
Each instance is configured with 18 vCPUs, 96 GB of system RAM and 900 GB of NVMe storage. Ubuntu 24.04 is provided with CUDA preinstalled, allowing customers to begin using the GPU environment immediately after deployment. The service also includes unlimited traffic under Contabo's fair use policy.
Pricing starts at €999, $1,199 or £969 per month, excluding VAT. Customers choosing longer-term commitments can receive a 15% discount on 12-month terms or a 20% discount on 24-month terms.
Additional options include Storage Extension and Private Networking, while supported 1-Click Apps include Ollama, n8n, OpenClaw and Hermes.
The GPU VPS is initially available in Contabo's EU and US Central locations, with capacity limited during the launch period. The service is designed for applications including AI inference, model fine-tuning, LoRA training, 3D rendering, compositing, computational fluid dynamics (CFD), finite element analysis (FEA/FEM), molecular dynamics and single-node data science workloads.
More information about Contabo GPU VPS is available on the
company's website.
