Before you start
Have wallet funds, a project if needed, and an SSH public key for shell access. Templates are filtered by GPU memory, hardware, and provider certification.
Steps
- Open GPU Cloud or Explore services and choose an available GPU offering. Review GPU count, VRAM, location, and rate.
- On Configure, open Deployment Template and choose a compatible template. Check its image, disk defaults, ports, and startup settings.
- Select the GPU count within the offering’s limit and provide enough container and workspace storage for models and data.
- Select your SSH public key, review the service name and project, and check the final order summary.
- Deploy once, then open the GPU Instance page. Wait until the container is ready before using Connections or SSH.
Check the result
The instance runs the chosen image and exposes the configured connections. Application readiness can take longer than provider provisioning.
Troubleshooting
If a template is missing, it may not be certified for that GPU or provider. Choose another compatible offering or use a custom image after checking its requirements. Do not assume every image supports every GPU.