Skip to content
AWS · Cost

AWS Lambda Pricing Calculator

Estimate your monthly AWS Lambda bill from memory, duration and requests, compare Arm with x86, add provisioned concurrency and find the cheapest memory size.

Last updated

Your workload
Examples:
Estimated cost — US East (N. Virginia)
$0.00
Per invocation—
GB-seconds / month0
ItemUsageRateCost

Right-size the memory

More memory also gives more CPU (one full vCPU at 1,769 MB), so CPU-bound functions often finish faster and can cost the same or less. Measure with AWS Lambda Power Tuning before changing production.

MemoryIf duration stays the sameIf CPU-bound (gets faster up to 1,769 MB)

Prices are AWS list prices for US East (N. Virginia), checked 20 October 2026 against the AWS Lambda pricing page. Other regions cost slightly more. Excludes data transfer, CloudWatch Logs, and services your function calls (API Gateway, SQS, DynamoDB …).

Frequently asked questions

How is AWS Lambda priced?

You pay $0.20 per million requests plus compute time, measured in GB-seconds: the memory you allocate (in GB) multiplied by how long each invocation runs, billed per millisecond. In US East, x86 costs $0.0000166667 per GB-second for the first 6 billion GB-seconds a month, and Arm (Graviton) costs $0.0000133334 for the first 7.5 billion, with lower rates above those volumes.

Is AWS Lambda free?

Every account gets 1 million requests and 400,000 GB-seconds of compute a month free, shared across all functions and both architectures. That covers, for example, 1 million invocations a month that each run for 400 ms with 1 GB of memory. The free tier does not apply to provisioned concurrency.

Is Arm (Graviton) cheaper than x86 on Lambda?

Arm compute is about 20% cheaper per GB-second, and AWS reports up to 34% better price-performance. Requests cost the same. Most Node.js, Python and Java functions run on Arm without changes; functions with native x86 dependencies need an Arm build.

What does provisioned concurrency cost?

Provisioned concurrency keeps environments initialised to avoid cold starts. You pay $0.0000041667 per GB-second (x86) for the time it is enabled, whether used or not, and invocations it serves are billed at a lower duration rate of $0.0000097222 per GB-second. It pays off when it is busy most of the time.

How can I reduce my Lambda bill?

Switch to Arm, right-size memory (test with AWS Lambda Power Tuning), cut duration by reusing connections outside the handler and avoiding waits on slow calls, batch messages from SQS or Kinesis, and set log retention on CloudWatch Logs, which is often a bigger cost than Lambda itself.

Practical guide

How AWS Lambda pricing works

Lambda bills two things: requests ($0.20 per million) and compute time in GB-seconds, which is the memory you allocate (in GB) multiplied by how long each invocation runs. Duration is billed per millisecond, rounded up. A 512 MB function running 200 ms uses 0.5 × 0.2 = 0.1 GB-seconds per call.

US East (N. Virginia) prices, checked 20 October 2026

  • x86: $0.0000166667 per GB-second for the first 6 billion GB-seconds a month, $0.0000150000 for the next 9 billion, and $0.0000133334 above 15 billion.
  • Arm (Graviton): $0.0000133334 for the first 7.5 billion, $0.0000120001 for the next 11.25 billion, and $0.0000106667 above 18.75 billion — about 20% cheaper.
  • Free tier: 1 million requests and 400,000 GB-seconds every month, shared across all functions in the account.
  • Provisioned concurrency: $0.0000041667 per GB-second while enabled (x86), and invocations it serves cost $0.0000097222 per GB-second. Arm: $0.0000033334 and $0.0000077778.
  • Ephemeral storage above the free 512 MB: $0.0000000309 per GB-second. Response streaming: $0.008 per GB after the first 100 GB.

What the calculator does not include

Data transfer, CloudWatch Logs (often larger than the Lambda bill — set a retention period), and the services your function calls, such as API Gateway, SQS, DynamoDB or Bedrock. Other Regions cost slightly more than US East.

Cutting the bill

  1. Switch to Arm if your dependencies support it.
  2. Right-size memory: CPU grows with memory (one full vCPU at 1,769 MB), so a CPU-bound function can finish faster and cost less with more memory. Measure with AWS Lambda Power Tuning.
  3. Reuse SDK clients and connections outside the handler, and avoid waiting on slow downstream calls.
  4. Batch SQS and Kinesis records, and filter events at the source so the function isn't invoked for nothing.
  5. For Amazon Connect flows, keep Lambda timeouts short (the flow waits up to 8 seconds) and cache lookups.

Official references

AWS AWS Bedrock Pricing Calculator AWS AWS Cron Expression Builder AWS ARN Parser & Builder AWS IAM Policy Generator Connect Amazon Connect Pricing Calculator AWS JWT & Base64 Decoder
Runs 100% in your browser. Your files and data never leave your machine. Nothing is uploaded or stored.

Need a hand with your AWS setup?

Get help from an AWS architect who builds and runs contact centre and AI workloads: IAM, data pipelines, Bedrock and cost reviews.

Talk to an architect