AWS Lambda Pricing in 2026
Plans, hidden costs, and alternatives compared
Is AWS Lambda worth the price?
Lambda pioneered serverless compute and remains the default choice in the AWS ecosystem.
The per-request pricing model is extraordinarily cost-effective for bursty, low-to-medium traffic workloads — a function handling 5 million requests/month at 256 MB with 200ms average duration costs about $2.10/month. The economics flip at high scale: sustained high-throughput workloads (>100M requests/month) often become cheaper on containers (ECS/Fargate) or even EC2.
Cold starts (100ms-2s depending on runtime) remain the main technical limitation. The pricing is transparent but has multiple dimensions (requests + duration + memory + optional provisioned concurrency) that make cost prediction harder than it looks.
Pricing Plans
Free Tier
$0
- 1M free requests/month
- 400,000 GB-seconds compute
- Never expires
- Included with AWS account
Pay-as-you-go
$0.20/million
- $0.20 per 1M requests
- $0.0000166667/GB-second (x86)
- Memory from 128 MB to 10 GB
- Billed per millisecond
Graviton2 (ARM)
$0.16/million
- 34% better price performance
- $0.0000133334/GB-second
- Same functionality as x86
- Recommended for cost savings
Savings Plans
Up to 17% off
- 1 or 3 year commitment
- Compute Savings Plans
- Flexible across regions
- Significant discount
Hidden Costs & Gotchas
Duration billing rounds up to the nearest millisecond, but memory billing is per-GB-second at allocated memory (not used memory). Allocating 1 GB but using 200 MB still bills for 1 GB — right-sizing memory is critical
Provisioned Concurrency
eliminates cold starts but charges $0.0000041667/GB-second even when idle. Provisioning 100 instances at 512 MB costs ~$54/month whether they handle traffic or not
Data transfer out
Lambda itself is cheap, but responses that cross AZ boundaries or leave AWS incur standard EC2 data transfer charges ($0.01-0.09/GB). High-traffic APIs serving large payloads can have transfer costs exceed compute costs
VPC-attached Lambda
connecting to RDS or ElastiCache inside a VPC adds ~200-500ms cold start overhead and creates ENI resources. VPC Lambda also requires NAT Gateway ($32/month + $0.045/GB) for outbound internet access
Ephemeral storage
default 512 MB included, up to 10 GB available at $0.0000000309/GB-second. Processing large files (video, ML models) can add meaningful storage costs
CloudWatch Logs
every Lambda invocation generates logs. At 5M invocations/month generating 1 KB each, log ingestion costs $2.50/month. High-verbosity logging on high-traffic functions can cost more than the compute itself
API Gateway costs are separate
if fronting Lambda with API Gateway, add $3.50/million requests (REST API) or $1.00/million (HTTP API). For some workloads, the gateway costs more than Lambda
Step Functions orchestration
coordinating multiple Lambda functions via Step Functions costs $25 per million state transitions — a 10-step workflow processing 1M executions/month costs $250/month in orchestration alone
How AWS Lambda Compares
API backend handling 10M requests/month, 256 MB memory, 200ms average duration, 12 months
Which Plan Do You Need?
1 million requests and 400,000 GB-seconds per month — free forever, not just 12 months. Covers most personal projects, webhook handlers, and cron jobs. At 128 MB memory, that is 3.2 million seconds of execution per month.
Pay $0.20 per million requests plus $0.0000166667 per GB-second. No servers to manage, no capacity planning. Scales from zero to thousands of concurrent executions. Best for event-driven architectures, API backends, and data processing pipelines.
34% better price-performance than x86 — $0.0000133334 per GB-second. Same Lambda experience, just runs on ARM. Most Node.js, Python, and Java workloads run without modification. The easiest optimization for existing Lambda functions.
Up to 17% discount on compute by committing to a consistent usage level. Stacks with Graviton2 for up to 45% total savings vs x86 on-demand. Only makes sense if you have 6+ months of usage data showing consistent patterns.
Our Recommendation
startup
Lambda free tier covers most early-stage workloads. Use HTTP API Gateway ($1/million requests) instead of REST API ($3.50/million) to cut gateway costs by 71%. Switch to Graviton2/ARM for an instant 20% cost reduction with no code changes for most runtimes. Set billing alarms at $10/month to catch runaway invocations early.
enterprise
Evaluate container alternatives (ECS Fargate, EKS) for functions running >100M invocations/month or >1 second average duration — the break-even point where containers become cheaper is typically 50-100M sustained invocations. Use Compute Savings Plans for predictable baseline + Lambda on-demand for burst. Right-size memory using Lambda Power Tuning (open-source tool). Budget for surrounding costs (API Gateway, CloudWatch, data transfer) — they often exceed Lambda compute costs at scale.
freelancer
Lambda free tier is one of the best deals in cloud computing for personal projects and client APIs. Pair with API Gateway HTTP APIs and DynamoDB free tier for a complete serverless backend at $0/month for low traffic. Monitor with CloudWatch free tier (5 GB logs, 10 custom metrics).
