Vultr (United States) is a cloud platform that bets on geographic coverage. It brings together compute instances, bare metal servers, managed Kubernetes, object storage and CDN.
What it does
- Ships applications close to users across many regions
- Runs GPU workloads and model training on demand
- Managed Kubernetes for teams that want less operations work
- Object storage and snapshots for backup
How it works
- Self-service account with panel, API and Terraform
- Hourly billing with a monthly cap per instance
- More than 30 locations across the Americas, Europe, Asia, Africa and Oceania
Public pricing
- Monthly fee per resource tier, with an hourly option
- Transfer and storage billed separately
- GPU and bare metal plans priced on their own
Strengths
- Many regions, including outside the main hubs
- Broad catalog, from VPS to GPU and bare metal
- Well-documented API and automation
Points of attention
- Performance and latency vary a lot by region
- Managed services are less deep than on large clouds
- Transfer costs deserve attention under heavy traffic