Compare Compute Engines

Evaluate our infrastructure options side-by-side. Filter by workload, performance tier, or pricing model to find your perfect match.

d
Specification
Standard VM Stable
General purpose virtual machines
From $0.012/hr
Container Engine Popular
Managed Kubernetes & ECS
From $0.025/hr
Functions
Event-driven compute
Pay per use
GPU Instances New
H100 / A100 accelerated
From $2.40/hr
Architecture AMD EPYC / Intel Xeon AMD EPYC Genoa ARM64 / x86_64 NVIDIA H100 / A100
Memory Config 2GB - 256GB 4GB - 512GB (Cluster) 128MB - 2GB per fn 80GB - 80GB VRAM
Storage NVMe SSD, Block Volumes Shared NFS, Ephemeral SSD Ephemeral only
Network I/O Up to 25 Gbps Up to 100 Gbps (Pod) Event-driven (WebSocket/HTTP) Up to 400 Gbps InfiniBand
Auto-Scaling Manual / CPU-based Horizontal & Cluster Autoscaler Instant cold/warm scale Queue-based scaling
Boot/Cold Start < 15 seconds < 30 seconds (node) 10-100ms (warm) < 45 seconds
OS Control Full SSH / Root Container images only Code only Full SSH / Root
Best For Monoliths, Legacy, Dev/Test Microservices, Modern Apps Webhooks, Cron, APIs LLMs, Training, Rendering
SLA Guarantee 99.99% Uptime 99.95% Uptime 99.9% Availability 99.9% Uptime
Get Started Launch VM Create Cluster Deploy Function Provision GPU

How to Choose?

Not sure which engine fits your architecture? Use these quick guidelines.

🚀 Need Maximum Control?

Go with Standard VMs. Full OS access, predictable pricing, and easy migration from on-premises setups.

📦 Building Microservices?

Container Engine handles orchestration, rolling updates, and service mesh without the operational overhead.

⚡ Want Zero Ops?

Serverless Functions scale to zero when idle. Perfect for background jobs, webhooks, and bursty traffic.

🤖 Training AI Models?

GPU Instances provide optimized NVIDIA drivers, CUDA toolkit pre-installed, and high-throughput networking.