2 × NVIDIA GeForce RTX 4090
Available
$0.70/h
≈ $0.35/GPU·h
total machine rate, per GPU equivalent second
Machine
- GPUs
- 2 × NVIDIA GeForce RTX 4090 (full-machine allocation)
- VRAM per GPU
- 24 GiB
- Aggregate VRAM
- 48 GiB
- RAM
- 128 GiB
- Disk
- 2000 GiB
- Region
- eu-central
- Measured bandwidth
- 940 Mbps
- Observed
- 2026-09-13 19:59 UTC (0 min ago)
Availability window
- From (UTC)
- 2026-09-13 00:00 UTC
- Until (UTC)
- 2026-09-20 00:00 UTC
Declared maintenance
- 2026-09-16 02:00 UTC → 2026-09-16 04:00 UTC — Kernel and driver update window (host-declared, versioned).
- Included disk100 GiB included in the rate
- $0.00
- Included egress10 GiB included in the rate
- $0.00
- Platform share (10%)Inside the displayed total rate — no post-stop surprise charges
- $0.00
- Total per machine-hour
- $0.70
Trust state
- Detectedhardware found
- RuntimeCUDA available
- Qualificationpassed
- Rentableyes
Qualified workloads: pytorch-train, pytorch-inference, jupyter. Memory class: dedicated.
Compatible templates: pytorch-cudajupyter-cuda
Machine manifest sha256:dddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd