Which is better, E2B or AWS Lambda MicroVMs?
It depends on where the rest of your stack lives. For agents that wait on a model, Runtime is the cheapest of the three Firecracker options: $0.64 for this job, against $2.76 on E2B and $4.20 on Lambda, and still the lowest with both CPUs busy. Pick E2B to self-host later under Apache-2.0 or to keep paused sandboxes with no expiry. Pick Lambda MicroVMs if everything else is on AWS and you want IAM, VPC access and many regions.
At a glance
| E2B | AWS Lambda MicroVMs | Runtime | |
|---|---|---|---|
| Isolation | Firecracker microVM, own kernel | Firecracker microVM | Firecracker microVM, own kernel |
| CPU | $0.0504 per allocated vCPU-hour | $0.0997 per baseline vCPU-hour, billed by second | $0.025 per vCPU-hour of measured CPU, with a small floor |
| Memory | $0.0162 per GiB-hour | $0.0132 per GB-hour, 2 GB per vCPU | $0.0075 per reserved GiB-hour |
| Largest | 8 vCPUs and 8 GiB | 4 vCPUs and 8 GB at peak | 16 vCPUs and 64 GiB, paid |
| Session | 1 hour on Hobby, 24 hours on Pro | State kept up to 8 hours | No time limit: runs while it works, or persistent |
| Idle | Pause keeps memory, with no expiry | Suspends into a snapshot after an idle time you set | Pause keeps memory while you have credit |
| Free start | $100 of usage credit, once | No free tier for MicroVMs | 100 free hours of a 1 GB machine in the first 30 days, no card |
Why is Lambda dearer than E2B here?
Lambda's memory rate is a little lower per unit, but its CPU is about twice E2B's: $0.0997 per vCPU-hour against $0.0504. For a 2 vCPU sandbox that is $0.199 an hour of CPU on Lambda and $0.101 on E2B, whatever the code does. Neither discounts a vCPU for sitting idle while the agent waits on its model. Runtime does: waiting costs a floor of a twentieth of a vCPU.
Cost for the same job
1,000 runs, 60 seconds each, 20 CPU-seconds of work, 2 vCPU and 4 GiB (a 4 GB baseline on Lambda):
- CPU
- 1,000 × 60 s / 3,600 × 2 × $0.0504
- $1.68
- Memory
- 1,000 × 60 s / 3,600 × 4 × $0.0162
- $1.08
- 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
At 100,000 runs a month that is $276.00 on E2B, $420.33 on Lambda MicroVMs and $63.89 on Runtime. If the agent kept both CPUs busy for the whole minute, the job would cost $1.33 on Runtime, still the lowest, with $2.76 on E2B and $4.20 on Lambda unchanged, since both already bill every vCPU as if it worked.
Sessions: one hour, eight hours, or as long as you like
E2B's free Hobby plan caps a session at an hour; Pro at $150 a month lifts it
to 24 hours. A Lambda MicroVM keeps its memory and disk for up to eight hours, and
suspends after an idle time you set, when compute stops and snapshot storage
starts at $0.08 per GB-month. A Runtime sandbox has no time limit: it runs while it
works, persistent: true keeps a paid one running even while idle, and a pause keeps its memory for as many days as you set, up to 365.
Which one to pick
- E2B if you want to self-host later (its runtime is Apache-2.0) or keep paused sandboxes with no expiry.
- Lambda MicroVMs if everything else is already on AWS: IAM, VPC access through a network connector, and many regions.
- Runtime for agents that wait on models: the lowest bill of the three at
any level of use, sandboxes up to 16 vCPUs and 64 GiB, and forks that copy a
running machine.
withruntime/e2bruns E2B code by changing one import.
See Runtime vs E2B and Runtime vs AWS Lambda MicroVMs.
Runtime Cloud
Runtime runs the same job for 77% less than E2B and 85% less than Lambda MicroVMs. 1,000 one-minute runs of a 2 vCPU, 4 GiB sandbox cost $2.76 on E2B, $4.20 on AWS Lambda MicroVMs and $0.64 on Runtime. All three use Firecracker; only Runtime bills the CPU the code used instead of every vCPU it holds.
Sign in with npx withruntime login and run the job on the free usage. E2B
code switches with one import line, as the migration guide
explains.
Sources
E2B rates checked 25 October 2026; Lambda rates checked 25 October 2026.
Facts on this page were checked on 26 October 2026.