# AWS Lambda MicroVMs

Official pricing: https://aws.amazon.com/lambda/pricing/  
Category: hyperscaler · Isolation: firecracker

## Pricing regimes (raw)

- **Arm64 (Graviton) baseline, us-east-1/us-east-2/us-west-2/ap-south-1** (resource): $0.0997/vCPU-h, $0.0132/GiB-h
- **Arm64 baseline, eu-north-1 (cheapest EU)** (resource): $0.09975/vCPU-h, $0.013208/GiB-h
- **x86 baseline, us-east-1/us-east-2/us-west-2/ap-south-1** (resource): $0.11756/vCPU-h, $0.015565/GiB-h [beta]
- **x86 baseline, eu-north-1** (resource): $0.117954/vCPU-h, $0.015617/GiB-h [beta]

## Features

Yes: snapshots, memory snapshots, pause/resume, persistent disk, idle auto-stop, custom image (Docker or snapshot), start from your own snapshot, your own Docker/OCI image, full VM (own kernel), root, HTTPS preview URLs, egress allowlist, open internet, static egress IP, private networking, SOC 2, HIPAA, SSO, arm64, HTTP method/path egress rules, wake on request, EU data residency, automatic snapshots, snapshots on demand, gVisor or VM (no shared kernel), inbound access rules, firewall inside (nftables)

No: fork/clone, volumes, ≥24 h sessions, Docker inside, nested virtualization, systemd, browser, desktop GUI, computer-use API, code interpreter, browser + desktop control, anti-bot stealth, CAPTCHA solving, residential IPs, preinstalled agents, SSH, public IPv4, custom domains, raw TCP inbound, self-hosting, BYOC, open source, GPU, Windows, macOS, Python, Node.js, GPU desktop, live resize, live fork (no pause), memory fork, MCP server, agent harness API, hosted agent API (their own agent), secret proxy, secret proxy for any API, extra volumes, shared volumes

Unknown: everything else. Evidence (source + quote) per feature: https://battleships.dev/data/providers/aws-lambda-microvms.json → feature_evidence

## Caveats

- All rates from the AWS Price List API (AWSLambda, publication 2026-09-19). Regions with MicroVM SKUs: us-east-1, us-east-2, us-west-2, eu-west-1, eu-north-1, eu-central-1, ap-south-1, ap-northeast-1, ap-southeast-1, ap-southeast-2.
- The knob rate_factor for burst is an approximation: burst is billed only for seconds it is consumed, at the same unit rates (per pricing page "baseline-plus-consumption").
- Snapshots (suspend = memory+disk): storage $0.0001111111/GB-h = $0.0811/GB-month; write $0.0037977/GB; read (on start/resume) $0.0015468/GB; MicroVM image storage has a 1-week minimum retention. Suspended VMs pay no compute.
- Included disk size, minimum billed duration and any per-launch fee are not published. Running disk: the AWS Price List API (AWSLambda us-east-1, fetched 2026-09-28) has only vCPU, memory and snapshot storage/read/write SKUs for MicroVMs, no running-disk meter, so disk_gib_month = 0 (disk of a running MicroVM is not billed separately; it is billed as snapshot storage once suspended). Maximum disk size per MicroVM is unpublished. Whether Compute Savings Plans or the Lambda free tier apply is not stated (assumed no).
- Max lifetime: the launch blog shows maximum-duration 28,800 s (8 h) as an example; the true cap is not stated.
- Identical to the announced AgentCore Runtime v2 committed baseline rates ($0.0997/vCPU-h, $0.0132/GB-h).
- Internet egress (AWS Price List API, AWSDataTransfer us-east-1, 2026-09): first 100 GB/month free aggregated across all services and regions, then $0.09/GB first 10 TB, $0.085 next 40 TB, $0.07 next 100 TB, $0.05 over 150 TB. eu-west-1 is also $0.09; ap-south-1 $0.1093; ap-northeast-1 $0.114; ap-southeast-1 $0.12 (first-10-TB tier). The engine uses the flat us-east-1 first tier.
- VPC access via Lambda Network Connector; NAT gateway gotcha: tasks/functions in private subnets reach the internet through a NAT gateway at $0.045/h per gateway (~$32.85/month per AZ) plus $0.045/GB processed (us-east-1; eu-north-1 $0.046, ap-south-1 $0.056), charged ON TOP of the $0.09/GB egress. Not modelled; public IPv4 ($0.005/h) is the usual alternative.
- Snapshot storage is region-priced: us-east-1/us-east-2/us-west-2 $0.0001111/GB-h ($0.0811/GB-mo), eu-north-1 $0.0001161 ($0.0848), ap-south-1 $0.0001267 ($0.0925), eu-west-1/eu-central-1 $0.0001322 ($0.0965), ap-northeast-1/ap-southeast-1/ap-southeast-2 $0.0001333 ($0.0973). The EU mode inherits the us-east-1 figure.
- Suspended state is kept for at most 8 hours ('you can suspend a running MicroVM for up to 8 hours'); suspendedDurationSeconds sets auto-termination. Per-MicroVM limits: concurrent connections 8 (1 vCPU) to 128 (16 vCPU); requests/s 40 (4 vCPU / 8 GB) to 160 (16 vCPU / 32 GB). SLA 99.5% (Lambda SLA). Shell access (not SSH) via SHELL_INGRESS connector and a shell auth token.

## How this provider charges


Stateful Firecracker microVMs (AWS's E2B/Daytona-style product). You configure a **baseline** (memory with a fixed
2 GB : 1 vCPU ratio; default 1 vCPU / 2 GB, max baseline 4 vCPU / 8 GB). Baseline is billed **per second while
running**; the VM bursts vertically up to 4x baseline (16 vCPU / 32 GB) and the extra is billed only for seconds
actually consumed ("baseline-plus-consumption"). Suspended VMs pay only snapshot storage. Rates below are from the
**AWS Price List API** (AWSLambda offer, publication 2026-09-19); the pricing page text describes only Graviton.

Reference 4 vCPU / 8 GB baseline: Arm us-east-1 = 4 x 0.0997 + 8 x 0.0132 = **$0.5044/h**; x86 = **$0.59476/h**.

## Regime table

| Regime | When it applies | How billed | Numbers (per vCPU-h / per GB-h) | Source |
|---|---|---|---|---|
| Arm baseline, us-east-1 / us-east-2 / us-west-2 / ap-south-1 | Running VM | Per second, allocated baseline | $0.0000276944/vCPU-s = $0.0997; $0.0000036667/GB-s = $0.0132 | Price List API Lambda-MicroVM-vCPU-Second-ARM |
| Arm baseline, eu-north-1 (cheapest EU) | Region pin EU | Same | $0.09975 / $0.013208 | Price List API EUN1 |
| Arm baseline, eu-west-1 / eu-central-1 | | Same | $0.10497 / $0.013897; $0.12599 / $0.016679 | Price List API |
| Arm baseline, ap-northeast-1 / ap-southeast-1 / ap-southeast-2 | | Same | $0.11607 / $0.015368; $0.12481 / $0.016524; $0.10079 / $0.013345 | Price List API |
| x86 baseline, us-east-1/2, us-west-2, ap-south-1 | x86 MicroVMs (SKU exists in API) | Same | $0.0000326557/vCPU-s = $0.11756; $0.0000043235/GB-s = $0.015565 | Price List API Lambda-MicroVM-vCPU-Second |
| x86 baseline, eu-north-1 | | Same | $0.117954 / $0.015617 | Price List API |
| Burst above baseline | Automatic, up to 4x baseline | Only seconds consumed, same unit rates | as above | https://aws.amazon.com/lambda/pricing/ |
| Suspended | `suspend` or auto-suspend (default after 5 min idle) | No compute; snapshot storage | $0.0001111111/GB-h = $0.0811/GB-month | Price List API Lambda-MicroVM-Snapshot-Storage-GB-Hour |
| Snapshot write / read | Each suspend (write) and start/resume (read) | Per GB | write $0.0037977138/GB; read $0.0015467699/GB | Price List API |
| MicroVM image storage | Custom images | Snapshot storage rate, **1-week minimum retention** | $0.0811/GB-month | pricing page |
| Egress / VPC | Outbound internet default on; Lambda Network Connector to VPC | Standard AWS data transfer | 100 GB/month free then $0.09/GB | AWSDataTransfer |
| Free tier / Savings Plans | — | Not stated | assumed none | — |

## Gotchas

1. **Baseline is billed on allocation, idle or not**; only the burst above baseline is usage-billed. A low baseline + burst is the cost lever, but burst capacity is best-effort.
2. **Snapshot I/O is metered**: an 8 GB suspend/resume cycle costs ~$0.043 in write+read fees on top of storage.
3. **eu-west-1 is 5% dearer than eu-north-1** and eu-central-1 is 26% dearer than us-east-1.
4. The "$0.0000291572/vCPU-s" rate quoted by a secondary source is the **eu-west-1 Arm** rate, not x86.
5. Max lifetime is configurable (launch example 28,800 s = 8 h); the true cap isn't published. Min billed duration and launch fee aren't published either.
6. Same unit rates as the announced AgentCore Runtime v2 committed baseline.

## Worked example

4 vCPU / 8 GB baseline x 50 concurrent x 8 h/day x 22 days = **8,800 VM-hours**, 30% CPU, 50 GiB snapshots
retained, 100 GiB egress (free tier).

| Regime | Calculation | Total / month |
|---|---|---|
| Arm, 4 vCPU/8 GB baseline, us-east-1 | 8,800 x 0.5044 = $4,438.72 + snapshots 50 x $0.0811 = $4.06 | **$4,442.78** |
| + snapshot I/O (suspend/resume 8 GB per VM-day) | 1,100 cycles x 8 GB x ($0.0037977 + $0.0015468) = $47.03 | **$4,489.81** |
| Arm, 1 vCPU/2 GB baseline + burst (optimistic: 0.2 vCPU avg burst, 6 GB burst memory held) | 8,800 x (0.1261 + 0.2 x 0.0997 + 6 x 0.0132) = 8,800 x 0.22524 = $1,982.11 + $4.06 | **~$1,986** |
| x86, 4/8 baseline, us-east-1 | 8,800 x 0.59476 = $5,233.89 + $4.06 | **$5,237.95** |
| Arm eu-north-1 | 8,800 x (4 x 0.09975 + 8 x 0.013208) = 8,800 x 0.504664 = $4,441.04 + $4.06 | **$4,445.10** |
| Anti-pattern: left running 24 h/day | 50 x 730 x 0.5044 = $18,410.60 | **$18,414.66** |