Accelerated computing
GPU Instances
NVIDIA RTX Pro Blackwell GPUs on dedicated NVMe instances. For training, inference, rendering, and HPC workloads. Access requires a quota increase — reach out to support@excloud.dev.
GPU lineup
All instances are dedicated, NVMe-backed. Each GPU family targets a different VRAM tier.
nv2a.xlarge
Entry-level GPU for prototyping, small model fine-tuning, and single-user inference.
nv3a.2xlarge
Balanced option for LoRA fine-tuning, batch inference, and medium-scale training runs.
nv1a.4xlarge
Maximum VRAM for large model training, full fine-tuning, and multi-user inference serving.
Getting access
Request quota
Email support@excloud.dev with your use case and desired GPU type. We review every request manually.
Get approved
Quota is granted based on availability. Once approved, GPU instances appear in your console and CLI.
Deploy & compute
Launch via exc compute create with the nv* instance type. NVMe storage is included with the instance.
Need GPU compute?
Tell us about your workload and we'll get you set up with the right GPU tier.