Amazon EC2 Pricing Calculator — Total Cost with Storage & Transfer
AWS EC2 cost estimator covering compute, EBS storage, and data transfer on one screen. Compare all EC2 pricing models — On-Demand, Reserved, Spot — for any instance type.
Pricing TLDR
- • On-Demand t3.micro starts at ~$0.0104/hr (~$7.50/mo) in us-east-1
- • Reserved Instances save up to 72% vs On-Demand for 3-yr all-upfront commitments
- • Spot Instances discount up to 90% for fault-tolerant workloads
- • EBS gp3 at $0.08/GB-month + data transfer from $0.09/GB outbound — the hidden costs most calculators miss
EC2 Cost Calculator — Compute, Storage, Data Transfer
Instance Configuration
Region
Operating System
Instance count
Hours per month (730h ≈ 100%)
Default 730h = always-on. Lower for dev/batch workloads.
Instance type
Selected: m5.large2 vCPU · 8 GiB · x86_64
| Type | Family | vCPU | Memory | Arch |
|---|---|---|---|---|
| t3.micropopular | General purpose | 2 | 1 GiB | x86 |
| t3.smallpopular | General purpose | 2 | 2 GiB | x86 |
| t3.mediumpopular | General purpose | 2 | 4 GiB | x86 |
| t3.largepopular | General purpose | 2 | 8 GiB | x86 |
| t4g.micropopular | General purpose | 2 | 1 GiB | ARM |
| t4g.smallpopular | General purpose | 2 | 2 GiB | ARM |
| t4g.mediumpopular | General purpose | 2 | 4 GiB | ARM |
| m5.largepopular | General purpose | 2 | 8 GiB | x86 |
| m5.xlargepopular | General purpose | 4 | 16 GiB | x86 |
| m6i.largepopular | General purpose | 2 | 8 GiB | x86 |
| m7i.largepopular | General purpose | 2 | 8 GiB | x86 |
| m7g.largepopular | General purpose | 2 | 8 GiB | ARM |
| c6i.largepopular | Compute optimized | 2 | 4 GiB | x86 |
| c7i.largepopular | Compute optimized | 2 | 4 GiB | x86 |
| c7g.largepopular | Compute optimized | 2 | 4 GiB | ARM |
| r6i.largepopular | Memory optimized | 2 | 16 GiB | x86 |
| r7i.largepopular | Memory optimized | 2 | 16 GiB | x86 |
| r7g.largepopular | Memory optimized | 2 | 16 GiB | ARM |
EBS Storage
Volume type
Size (GB)
Provisioned IOPS (3,000 free)
Throughput (MB/s) (125 free)
Data Transfer
Internet egress (GB/month)
First 100 GB free
Inter-AZ (GB/month)
$0.01/GB each direction
Inter-region (GB/month)
$0.02/GB outbound
Estimated monthly cost
$114.08
Compute (1× m5.large)
$70.08
EBS (100 GB gp3)
$8.00
Internet egress (500 GB)
$36.00
Effective compute rate: $0.096/hr
US East (N. Virginia) · effective $0.096/hr OD
Pricing Model Comparison — On-Demand vs Reserved
Click any row to apply that commitment to the total above. Savings shown vs On-Demand for the current instance (m5.large, 1×, linux).
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Commitment
Effective $/hr
Upfront
Monthly
Savings
Tired of AWS billing surprises?
Track your actual AWS spending as it happens. Set custom budget alerts from your menubar.
Privacy-first desktop app. No sign-up required.
About Amazon EC2
What is Amazon EC2?
Amazon Elastic Compute Cloud (EC2) is AWS's virtual server offering — on-demand Linux and Windows VMs across 30+ regions. EC2 spans 800+ instance types organized into five families: General Purpose (t/m), Compute Optimized (c), Memory Optimized (r/x), Storage Optimized (i/d), and Accelerated Computing (p/g/inf/trn for GPU and AI). Pricing is per-second (60-second minimum) and varies by instance type, OS, region, and commitment model (On-Demand, Reserved, Savings Plans, Spot).
- Five Instance Families by Workload: General Purpose (t3/t4g burstable, m5/m6i/m7i balanced) for web servers and dev environments. Compute Optimized (c5/c6i/c7i) for CPU-bound workloads like batch processing and gaming. Memory Optimized (r5/r6i/r7i, x1/x2) for in-memory databases and analytics. Storage Optimized (i3/i4i NVMe) for databases needing local NVMe. Accelerated (g5/g6 GPU, p4/p5 H100, inf2/trn1 for AI/ML) for inference and training.
- Graviton (ARM) vs x86: AWS's Graviton processors (t4g, m7g, c7g, r7g, etc.) deliver up to 40% better price-performance than equivalent x86 instances. ARM instances are ~10-20% cheaper list price and more energy-efficient. Migration requires a one-time recompile — most mainstream languages (Go, Java, Python, Node.js) run on ARM with zero code changes. Graviton is the default cost-optimization win if your stack supports it.
- Pricing Models: OD, RI, SP, Spot: On-Demand: pay per second, no commitment, full price. Reserved Instances: 1-year or 3-year commit, save 37-72%. Savings Plans: the modern alternative to RIs — same discounts but apply across instance families, regions, and even to Lambda/Fargate; more flexible. Spot Instances: up to 90% off but can be reclaimed with 2 minutes notice, for batch / fault-tolerant workloads. Dedicated Hosts: entire physical server for licensing compliance.
When to Use Amazon EC2
Pick EC2 when you need full OS control, custom software stacks, or long-running workloads. For simple serverless workloads consider Lambda (cheaper for low traffic, simpler ops). For all-inclusive low-cost VPS without the AWS sprawl, consider Hetzner Cloud.
Ideal for
- Production web/API servers with predictable load (Reserved Instances or Savings Plans)
- Batch and ML training workloads with Spot Instances for 70-90% savings
- Legacy applications that require specific OS/kernel versions
- GPU-accelerated inference and training (g5, p5, trn1 instance families)
- Database and analytics workloads needing high memory or local NVMe storage
Not ideal for
- Event-driven workloads with infrequent traffic (AWS Lambda is cheaper)
- Static websites and simple container hosting (App Runner, Fargate, or a VPS may be better)
- Teams that want fully-managed infra with no OS patching (consider RDS, Aurora, App Runner)
- Budget-sensitive single-server workloads with EU-only latency requirements (Hetzner is 5-10× cheaper)
Amazon EC2 Pricing Breakdown
Free Tier
EC2 has a 12-month Free Tier for new AWS accounts. After 12 months, all usage is billed at standard rates — there is no permanent free tier for EC2.
- 750 hours/month of t2.micro or t3.micro (Linux or Windows) — enough to run one instance 24/7
- 30 GB of EBS gp2/gp3 storage (combined across volumes)
- 2 million I/O operations on EBS
- 1 GB of snapshot storage
- 100 GB of outbound data transfer to the internet (also applies after free tier period ends)
- Free tier applies globally, not per-region
EC2 Pricing Models Explained
On-Demand — pay per second, no commit
Full price, 60-second minimum. Best for unpredictable, short-term, or dev/test workloads. Use as the baseline to compare against Reserved / Savings Plans / Spot.
Reserved Instances — 1 or 3 year commit, 37-72% savings
Three payment options: No Upfront, Partial Upfront, All Upfront (cheapest). Two classes: Standard (locked to instance family) and Convertible (can swap families, smaller discount). All-Upfront 3-year Standard gives the deepest discount, usually 65-72% off On-Demand.
Savings Plans — flexible RI alternative (recommended default for steady workloads)
Compute Savings Plans: commit to a $/hr spend for 1 or 3 years, apply automatically to EC2, Lambda, and Fargate across any instance family and region. EC2 Instance Savings Plans: similar discount to RIs but with modern flexibility. Usually the best default for steady-state workloads — more flexible than RIs with similar savings.
Spot Instances — up to 90% off, but interruptible
Bid for spare capacity. Prices fluctuate but are usually 60-90% below On-Demand. Instances can be reclaimed with 2 minutes warning. Best for batch processing, CI/CD, ML training, fault-tolerant workers. Modern tooling (Spot Fleet, Karpenter) handles interruptions automatically.
Hidden Costs Most Calculators Miss
Data transfer (egress)
$0.09/GB for first 10TB/month outbound to the internet after 100GB free — easily $90+/month for a modest API. Inter-AZ traffic is $0.01/GB each direction (~$20/GB/month at 1TB bi-directional). Inter-region traffic is $0.02/GB outbound.
EBS storage and IOPS
gp3 at $0.08/GB-month, plus $0.005/IOPS-month beyond the 3,000 free IOPS and $40.96/MBps-month beyond 125 free MB/s. io2 is $0.125/GB-month plus $0.065/IOPS-month. A 500GB gp3 volume with 10,000 IOPS and 500 MB/s: $40 + $35 + $15,360 = ~$15K/month (use io2 Block Express instead at this level).
Elastic IPs
Elastic IPs attached to running instances are free. But since Feb 2024, all public IPv4 addresses cost $0.005/hour (~$3.60/month each). Unused Elastic IPs are still billable — release them.
Stopped instance zombies
Stopping an EC2 instance pauses compute charges but EBS volumes, snapshots, and Elastic IPs keep billing. 'Zombie' stopped instances accrue $8-50/month each silently. CostGoat surfaces these in your monitoring.
5 Amazon EC2 Cost Optimization Tips
Switch to Graviton (ARM) for 20-40% savings
Most mainstream languages (Go, Java, Python, Node.js, Rust) run on AWS Graviton ARM with zero code changes — just a container rebuild. m7g.large is ~18% cheaper than m7i.large with similar performance, and r7g.xlarge is up to 40% better price-performance for memory-heavy workloads. Graviton is the single highest-leverage cost optimization for most teams.
Use Compute Savings Plans before RIs
Compute Savings Plans give ~same discount as Standard RIs but apply flexibly across instance families, regions, and even Lambda/Fargate. For any steady-state workload (>60% utilization), commit to a 1-year or 3-year Compute SP and stop worrying about right-sizing across commitment families. RIs only make sense if you truly lock an instance type for 3 years.
Run batch workloads on Spot
Spot Instances discount 60-90% off On-Demand. They can be reclaimed with 2 minutes notice — fine for CI runners, data processing, ML training, and stateless workers. Use Spot Fleet or Karpenter to handle interruptions automatically. Typical savings: a c7i.xlarge batch job that costs $122/month on OD runs ~$15-25/month on Spot.
Rightsize before you commit
AWS Compute Optimizer and CloudWatch both surface underutilized instances. An m5.2xlarge running at 15% CPU for 6 months is a $400/month waste — downsize to m5.large ($70/month) and save $330/month. Don't commit RIs/SPs until you've rightsized — otherwise you're locking in waste.
Track stopped instances and orphaned EBS volumes with CostGoat
Stopped EC2 instances don't bill for compute but keep billing for their EBS volumes ($8-50/month each). Orphaned Elastic IPs bill at $3.60/month. CostGoat surfaces these zombie costs across your entire AWS account in real-time, so you catch them the day they appear — not on next month's bill.
Real-World EC2 Cost Examples
Total cost (compute + storage + transfer) for common EC2 workloads, all in us-east-1 unless noted.
Small web app
~$54/mo
• 1× t3.small Linux On-Demand (~$15/mo)
• 30 GB gp3 storage (~$2.40/mo)
• 500 GB/month egress (~$36/mo)
• Switch to t4g.small (ARM): ~$12/mo compute
• Or move to Hetzner CX23 for ~$4/mo
Personal blog or SaaS MVP — single instance with moderate egress.
Production web tier
~$675/mo OD · ~$538/mo RI
• 3× m6i.large Linux On-Demand (~$210/mo)
• 100 GB gp3 each (~$24/mo total)
• 5 TB/month egress (~$441/mo)
• 3-year All-Upfront Standard RI: ~$73/mo compute (+$3,800 upfront)
• Savings Plan recommended for flexibility
Mid-traffic SaaS behind an ALB with 3 instances and 5TB/month egress.
Self-managed database
~$1,140/mo
• 1× r6i.2xlarge Linux On-Demand (~$368/mo)
• 500 GB io2 + 10,000 IOPS (~$712/mo)
• Inter-AZ replication 100 GB/day (~$60/mo)
• RDS/Aurora may be cheaper once ops overhead is counted
Production Postgres on EC2 with high-IOPS io2 storage and HA replica.
GPU inference (Spot)
~$220-310/mo on Spot
• 1× g5.xlarge (A10G GPU)
• On-Demand baseline: ~$732/mo
• Spot discount 60-70%: ~$220-310/mo
• 200 GB gp3 for model weights (~$16/mo)
• Use SageMaker endpoints for always-on with auto-scale
LLM inference server for a prototype running Llama 70B on a single GPU.
Always-on dev server
~$29/mo OD · ~$14/mo RI
• t4g.medium (ARM, 2 vCPU, 4 GiB)
• 50 GB gp3 storage (~$4/mo)
• 3-year All-Upfront Standard RI: ~$10/mo compute
• Comparable Hetzner CAX21 ~€8/mo if EU latency works
Team shared dev environment running 24/7 on cheap ARM Graviton.
CI/CD runner fleet (Spot)
~$29/mo on Spot
• Fleet of Spot c7g.xlarge (ARM)
• ~4 concurrent runners × 6 hrs/day avg
• Spot ~$0.04/hr vs OD $0.145/hr
• On-Demand equivalent would be ~$100/mo
• Interruptions handled by re-queueing jobs
Self-hosted GitHub Actions runners for a 20-engineer team.
Track Your Cloud Costs in Real-Time
Monitor cloud spending across all your providers and APIs from one place. Set budget thresholds and get alerts when costs exceed your limits.
Privacy-first desktop app. 7-day free trial, no sign-up required.
Amazon EC2 Pricing FAQ
Common questions about AWS EC2 costs, Reserved Instances, stopped instance billing, and the cheapest instance types.
