AWS Lambda Cost Estimator
Estimate monthly and yearly Lambda cost from requests, runtime duration, memory, architecture, and optional extra ephemeral storage.
How AWS Lambda pricing works
AWS Lambda pricing is usually simple once you break it into pieces. Most teams pay for three things: requests, compute duration, and (if used) extra ephemeral storage. This calculator focuses on those components so you can quickly estimate cost before shipping a serverless workload.
- Requests: Charged per million invocations (after free tier, if applicable).
- Duration: Charged in GB-seconds, based on memory size and execution time.
- Ephemeral storage: Additional charge only for storage above 512 MB.
The core Lambda cost formula
The most important metric is GB-seconds. You get it by multiplying invocation count by duration (seconds) and memory (GB):
GB-seconds = Requests × (Duration ms ÷ 1000) × (Memory MB ÷ 1024)
Then multiply billable GB-seconds by your region + architecture rate. Arm/Graviton is typically cheaper than x86, so if your runtime is compatible, architecture can be a direct cost optimization lever.
Free tier impact
Lambda includes a monthly free tier (for eligible accounts): 1 million requests and 400,000 GB-seconds. For low-to-moderate traffic workloads, this can dramatically reduce costs. In this calculator, you can turn free tier on or off to model both startup-stage and mature production scenarios.
How to use this AWS Lambda pricing calculator
- Choose your region and architecture (x86 or Arm).
- Enter monthly invocations.
- Enter realistic average duration in milliseconds.
- Set memory size (MB).
- Optionally enter ephemeral storage above 512 MB.
- Decide whether to apply free tier and click calculate.
The output includes a full breakdown: total GB-seconds, billable requests, request cost, compute cost, ephemeral storage cost, monthly total, and yearly estimate.
Example scenarios
1) Lightweight API endpoint
Suppose you run 3 million requests/month at 128 MB and 50 ms average duration. Your GB-seconds are small, and free tier may cover a large share of usage. In many cases, the monthly bill can be surprisingly low.
2) Data processing function
If you process 20 million events/month at 1024 MB and 900 ms average duration, compute dominates cost. Here, optimizing execution time and memory tuning can save much more than reducing invocation count alone.
3) Image/video transformation
Workloads with large temporary files may require extra ephemeral storage. In those cases, include storage above 512 MB in your estimate so you capture the full monthly spend.
Ways to reduce Lambda cost
- Profile memory settings: More memory can reduce runtime; cheapest option is not always lowest MB.
- Optimize cold starts: Smaller packages and efficient initialization reduce billed time.
- Use Arm where possible: Graviton pricing is often lower than x86.
- Batch event processing: Fewer invocations can reduce request charges.
- Trim retries and noisy traffic: Unexpected retries can silently inflate usage.
Common estimation mistakes
- Using peak duration instead of realistic average duration.
- Forgetting free tier assumptions when forecasting production growth.
- Ignoring architecture differences (x86 vs Arm).
- Not accounting for additional ephemeral storage in media/data workloads.
- Assuming one region’s rates apply globally.
Final notes
This page is designed for planning and quick what-if analysis. AWS pricing can change over time and varies by region, so always verify final numbers with official AWS pricing documentation before making budgeting or contractual decisions.