AWS Lambda cost calculator: estimate your monthly bill

This AWS Lambda cost calculator estimates your monthly bill from three numbers you already know: requests per month, average duration and memory. It uses the current US East (N. Virginia) prices from the AWS price list published on 2026-10-16, covers x86 and Arm, the free tier, volume discounts, extra ephemeral storage and provisioned concurrency, and it reproduces AWS's own pricing examples to the cent.

Storing data in DynamoDB too? DynamoDB cost calculator →

Calculator

The defaults are AWS's mobile-backend example: 3 million requests, 120 ms, 1,536 MB, which comes to $2.73 a month. Change any value and the estimate updates.

How Lambda pricing works

1. 1
   Requests. Every invocation costs $0.20 per million, whatever it does.
2. 2
   Duration in GB-seconds. Lambda bills the time your code runs, rounded up to the nearest millisecond, multiplied by the memory you configured. 3 million requests of 120 ms at 1.5 GB is 3,000,000 x 0.12 x 1.5 = 540,000 GB-seconds.
3. 3
   Architecture. Arm (Graviton) functions pay a lower price per GB-second than x86. The request price is the same.
4. 4
   Volume tiers. Very large accounts pay less per GB-second after 6 billion GB-seconds a month on x86, or 7.5 billion on Arm.
5. 5
   Extras. Ephemeral storage above the free 512 MB, and provisioned concurrency if you keep environments warm.

Current prices (US East)

From the AWS Price List for US East (N. Virginia), published 2026-10-16. Other regions differ slightly.

Worked examples

Both match AWS's published examples, and both are in the calculator's test suite.

The free tier

Every month, 1 million requests and 400,000 GB-seconds of compute are free, shared across x86 and Arm functions in the account. That is enough for about 3.2 million seconds at 128 MB, or 400,000 seconds at 1 GB. The free tier does not cover provisioned concurrency.

Costs the calculator leaves out

Lambda is rarely the only line on the bill for a serverless API. Check these too:

• API Gateway. Requests through API Gateway are billed separately. A Lambda function URL has no extra charge, which suits simple webhooks and internal APIs.
• CloudWatch Logs. Every console.log is ingested and stored. Chatty logging can cost more than the function itself.
• Data transfer. Data sent out to the internet, or across regions, is billed per GB.
• What the function calls. DynamoDB, S3, SQS and third-party APIs all bill on their own. The DynamoDB cost calculator covers the most common one.

How to reduce your Lambda bill

1. 1
   Right-size memory. More memory also means more CPU, so a function can finish faster and cost less at a higher setting. Measure rather than guess; the open-source AWS Lambda Power Tuning tool runs your function at several sizes and charts cost against speed.
2. 2
   Move to Arm. Most Node.js and Python functions run on Arm without changes, for 20% less per GB-second.
3. 3
   Stop paying to wait. Time spent waiting on a slow API is billed like computation. Set tight timeouts, run independent calls in parallel, and hand slow work to a queue.
4. 4
   Batch. Processing 100 SQS messages per invocation instead of one cuts request charges and start-up overhead.
5. 5
   Use provisioned concurrency only for steady, latency-sensitive traffic. It is billed for every hour it is on, used or not.
6. 6
   Trim logs. Log at info level in production and set a retention period on log groups.

Common questions

Want a second opinion on your AWS bill?

I build and tune serverless backends on AWS: right-sized functions, sensible queues and APIs that do not cost more than they earn. See AWS serverless development.