Vorschaubeispiel 8: 4× RTX PRO 6000 — Open-LLM sofort nutzbar (gpt-oss-120b, Mistral Small 4 · vLLM), ab 12 h buchbar (Beispiel-Slot)
EU sovereign · Data centre in Germany
The provider is registered in the EU, is not controlled by a company outside the EU and operates the hardware in an EU data centre. It is therefore not bound by disclosure orders under the US CLOUD Act. This is a requirement for every provider on CMB (Terms, clause 5f).
The provider is registered in the EU, is not controlled by a company outside the EU and operates the hardware in an EU data centre. It is therefore not bound by disclosure orders under the US CLOUD Act. This is a requirement for every provider on CMB (Terms, clause 5f).
Lot details
| Provider | |
| GPU type | RTX PRO 6000 × 4 |
| Access types | SSH (VM/bare metal) Containers (Docker/K8s) |
| Proof of delivery | Hourly Hourly principle — only delivered hours are billed, spec check every hour |
| Hardware & environment | 96 GB VRAM/GPU · 512 GB RAM · 64 CPU-cores (AMD EPYC 9354) · 4000 GB NVMe · 10000 Mbit/s · Ubuntu 22.04 · CUDA 12 · PyTorch 2.x |
| Open-weights models | Mistral Small 4, Qwen3.8 27B, gpt-oss-120b, Llama 3.3 70B Served via: vLLM (OpenAI-kompatible API) |
| Fits | AI models up to approx. 280 B parameters (Int8) · 560 B (Int4) 384 GB VRAM total · FP16 up to approx. 140 B · Guide value for inference (running a trained model), not for training. |
| Location | Germany · Frankfurt EU sovereign |
| Available (from–to) | 10.11.2026 00:00 bis 10.12.2026 00:00 (720 Stunden) |
| Booking | period of your choice: from 12 h · up to the whole window · 4 GPU — The provider sets minimum and maximum length. |
| Available | 22.11. 00:00 – 10.12. 00:00 (432 h) |
| Pricing | Slots · pick your period fixed price per GPU-hour · book instantly, no bidding |
| Fixed price | 1,10 €/GPU-h net≈ 184,80 €/GPU-week |
| Status | active — bookable for: |
Book a period
Fixed price: 1,10 € /GPU-h net
You choose start and end yourself — from 12 h · up to the whole window, 4 GPU.
No bidding: the booking is binding immediately.
Bookings (0)
No booking yet.
Match with request: Matching compute requests (0)
Automatic match on GPU type, count, delivery window, price, hardware, certifications and access type. Only matches from 50 % — from running auctions and open requests.
No matching entries at the moment.
