Skip to main content
In this guide, we’ll dive into AWS Lambda pricing—covering the free tier, core billing components (requests and gigabyte-seconds), memory/CPU allocation, architecture choices, and additional cost factors like storage and concurrency. By understanding these elements, you can optimize your serverless workloads for both performance and cost.

​
Free Tier

AWS Lambda incurs no upfront fees. Each month, you receive:
  • 1 million free requests
  • 400 000 GB-seconds of compute time
Free-tier usage resets every month. Usage beyond these amounts is billed according to the rates in your AWS Region.

​
Core Pricing Components

Lambda charges are based on:
  1. Number of requests
  2. Compute time measured in gigabyte-seconds (GB-s)
The image outlines the components of Lambda cost, highlighting the number of requests and gigabit seconds (amount of time multiplied by the amount of resources).

​
1. Requests

Each time your function handler begins execution—triggered by API Gateway, event sources, or manual tests—it counts as one request.
Price = (Total requests) × (Request rate per million)

​
2. Gigabyte-Seconds

Compute time is billed in GB-s, calculated as: 
GB-s = (Allocated memory in GB) × (Execution duration in seconds)
  • Timer starts when the handler is invoked and stops after shutdown code completes.
  • Rounding is to the nearest millisecond.
AWS rounds up to the nearest ms, so a 1.0001 ms function is billed as 2 ms.
The image is a diagram illustrating the flow of data from a source to AWS Lambda, where a Lambda function is executed, and then to another AWS service. It includes a timeline for initialization and shutdown, labeled "Gigabit Seconds."

​
Memory and CPU Allocation

When you configure a Lambda function, you select memory between 128 MB and 10 240 MB. CPU power scales linearly with memory—more RAM also means more CPU share, speeding up execution and reducing GB-s.
The image illustrates a concept related to "Gigabit Seconds" with a focus on CPU and memory allocation for a Lambda function, showing a range from 128 MB to 10,240 MB.

​
CPU Architectures

Lambda supports two processor types. Choosing the right architecture can yield significant savings:
ArchitecturePrice-Performance ImprovementBest for
x86_64BaselineGeneral workloads
ARM64 (Graviton2)Up to 34% betterCost-sensitive, high-throughput jobs
Tip: ARM64 functions may require compatibility testing for native dependencies.

​
Additional Cost Factors

Besides requests and GB-s, consider these extras:
ComponentDescriptionBilling Unit
Ephemeral storageScratch space (/tmp) from 512 MB to 10 GBPer GB-month (region-based)
Provisioned concurrencyKeeps functions warm to avoid cold startsPer concurrency-second
Provisioned concurrency charges apply even during idle periods. Use only for latency-sensitive functions.
The image outlines pricing components for a service, highlighting memory range, ephemeral storage, provisioned concurrency, and cold start avoidance. It includes region-specific pricing and uses a lambda symbol.

​
Summary

Your AWS Lambda costs depend on:
  • Free-tier utilization
  • Request count
  • GB-seconds (memory × duration)
  • Memory/CPU allocation
  • CPU architecture (x86_64 vs ARM64)
  • Additional storage and concurrency settings

​
Next Steps

Use the AWS Pricing Calculator to model workloads and forecast costs.

​
References

Watch Video