Which is better, Vercel Sandbox or AWS Lambda MicroVMs?
It depends on which cloud your app lives in. For agents that mostly wait on a model, Runtime bills active CPU as Vercel does, at a lower rate. The benchmark costs $0.64 on Runtime, against $2.12 on Vercel and $4.20 on Lambda. Runtime keeps memory across pauses, and it runs Vercel Sandbox code after one import change. Pick Vercel Sandbox for an app hosted on Vercel, or for a choice of 19 regions. Pick Lambda MicroVMs for workloads that must stay inside an AWS VPC.
At a glance
| Vercel Sandbox | AWS Lambda MicroVMs | Runtime | |
|---|---|---|---|
| Isolation | Firecracker microVM, own kernel | Firecracker microVM | Firecracker microVM, own kernel |
| CPU | $0.128 per active CPU-hour, iad1 |
$0.0997 per baseline vCPU-hour | $0.025 per vCPU-hour of measured CPU, with a small floor |
| Memory | $0.0212 per provisioned GB-hour | $0.0132 per GB-hour | $0.0075 per reserved GiB-hour |
| Largest | 8 vCPUs and 16 GB on Pro; 32 and 64 GB on Enterprise | A 4 vCPU, 8 GB baseline | 16 vCPUs and 64 GiB, paid |
| Idle | Stop keeps the filesystem; processes start again | Suspend keeps memory and disk, up to 8 hours | Pause keeps memory and processes while you have credit |
| Regions | 19 | AWS regions | One US region |
Active CPU against the baseline
Vercel and Runtime both charge CPU only while it works. That is why Vercel beats Lambda on a waiting agent, even though its CPU rate is the highest of the three. Lambda charges its baseline for every second a MicroVM is up. The order flips when the work gets busy, because then Vercel's $0.128 rate bites.
Cost for the same job
1,000 runs of 60 seconds on 2 vCPU and 4 GiB (4 GB on both rivals), 20 CPU-seconds of work each:
- CPU
- 1,000 × 20 s / 3,600 × $0.128
- $0.71
- Memory
- 1,000 × 60 s / 3,600 × 4 × $0.0212
- $1.41
- CPU
- 1,000 × 60 s × 2 × $0.0000276944
- $3.32
- Memory
- 1,000 × 60 s × 4 × $0.0000036667
- $0.88
- CPU
- 1,000 × 20 s / 3,600 × $0.025
- $0.14
- Memory
- 1,000 × 60 s / 3,600 × 4 × $0.0075
- $0.50
That is $212.44, $420.33 and $63.89 for a month of 100,000 runs. With both CPUs busy for the whole minute, the job costs $5.68 on Vercel, $4.20 on Lambda and $1.33 on Runtime. Lambda becomes cheaper than Vercel, and Runtime stays cheapest.
Which one to pick
- Vercel Sandbox for apps already on Vercel, in any of 19 regions.
- Lambda MicroVMs for workloads that live inside an AWS VPC.
- Runtime for active-CPU billing at 80% below Vercel's rate, memory kept
across pauses, and
withruntime/vercelfor Vercel Sandbox code after one import change.
See Runtime vs Vercel Sandbox and Runtime vs AWS Lambda MicroVMs.
Runtime Cloud
Runtime runs the benchmark for 70% less than Vercel and 85% less than Lambda. The benchmark is 1,000 one-minute runs of a 2 vCPU, 4 GiB sandbox that computes for 20 CPU-seconds of each minute: $2.12 on Vercel Sandbox, $4.20 on AWS Lambda MicroVMs, $0.64 on Runtime. All three run Firecracker.
Start free with npx withruntime login. For Vercel Sandbox or Lambda code, the
migration guide lists what changes.
Sources
Vercel rates checked 25 October 2026; Lambda rates checked 25 October 2026.
- Vercel Sandbox pricing, Understanding Vercel Sandboxes
- AWS Lambda pricing, Lambda MicroVMs
- Runtime pricing
Facts on this page were checked on 26 October 2026.