E2B alternatives, measured
Looking for an alternative to E2B? We timed E2B and 8 other cloud sandbox products through one SDK, on two tracks: the same 4 vCPU, 8 GB machine for every provider, and each provider's default machine. Below, each alternative is set against E2B test by test: how many tests it is ahead on, behind on or tied, its cold start and tool replay next to E2B's, its cost per task and whether it can pause. Two medians inside each other's 95% intervals count as a tie, not a win.
Sandbox bench runs on Sandbox SDK, which the same team (opencoredev) builds. We may work with some of these providers. No provider paid for these results or decided how they were tested.
E2B alternatives on the same 4 vCPU, 8 GB machine
Every provider asked for Boat's machine, 4 vCPU and 8 GB, or its nearest size. Runs from 2026-10-22 to 2026-10-23, 9 providers. Of the 8 alternatives, 3 start faster than E2B and 5 slower, none tied within the 95% intervals. Ranks are out of every provider on this track; see them all on the full results.
E2B on this track, for reference
- Cold start, median
- 0.79 s 4th of 9
- Tool replay, median
- 15 s 3rd of 9
- Cost per 1,000 tasks
- $1.56 (metered)
- Pause and resume
- pause 1.29 s, resume 0.51 s (median of 100)
A task is: create a sandbox, run the 100-call tool replay, delete it. Cost per task comes from a separate cost run, billed per sandbox. Cost kinds differ between providers (billed, metered, estimate) and aren't directly comparable.
Freestyle
Ahead of E2B on 9 of the 24 tests both finished, behind on 13, tied on 2.
- Cold start, median
- 0.69 s 3rd of 9ahead of E2B
- Tool replay, median
- 24 s 6th of 9behind E2B
- Cost per 1,000 tasks
- $1.97 (estimate)
- Pause and resume
- pause 0.32 s, resume 0.35 s (median of 100)
Tenki
Ahead of E2B on 7 of the 24 tests both finished, behind on 16, tied on 1.
- Cold start, median
- 1.1 s 5th of 9behind E2B
- Tool replay, median
- 21 s 4th of 9behind E2B
- Cost per 1,000 tasks
- $1.79 (estimate)
- Pause and resume
- pause 60.53 s, resume 2.45 s (median of 77)
boxd
Ahead of E2B on 8 of the 24 tests both finished, behind on 16, tied on 0. Machine differs: it measured something other than the requested size inside.
- Cold start, median
- 1.2 s 7th of 9behind E2B
- Tool replay, median
- 34 s 8th of 9behind E2B
- Cost per 1,000 tasks
- $2.00 (metered)
- Pause and resume
- pause 0.12 s, resume 0.40 s (median of 99)
Superserve
Ahead of E2B on 19 of the 24 tests both finished, behind on 5, tied on 0.
- Cold start, median
- 0.52 s 1st of 9ahead of E2B
- Tool replay, median
- 13 s 2nd of 9ahead of E2B
- Cost per 1,000 tasks
- $1.32 (billed)
- Pause and resume
- pause 0.39 s, resume 0.53 s (median of 100)
Upstash Box
Ahead of E2B on 9 of the 24 tests both finished, behind on 14, tied on 1.
- Cold start, median
- 2.7 s 9th of 9behind E2B
- Tool replay, median
- 27 s 7th of 9behind E2B
- Cost per 1,000 tasks
- $0.34 (billed)
- Pause and resume
- pause 0.84 s, resume 0.57 s (median of 99)
Daytona
Ahead of E2B on 13 of the 23 tests both finished, behind on 9, tied on 1.
- Cold start, median
- 1.3 s 8th of 9behind E2B
- Tool replay, median
- 12 s 1st of 9ahead of E2B
- Cost per 1,000 tasks
- $1.36 (billed)
- Pause and resume
- Daytona's container sandboxes have no pause: stop and start restarts the sandbox and ends every running process.
Upstash Ephemeral Box
Ahead of E2B on 10 of the 23 tests both finished, behind on 11, tied on 2.
- Cold start, median
- 0.68 s 2nd of 9ahead of E2B
- Tool replay, median
- 22 s 5th of 9behind E2B
- Cost per 1,000 tasks
- $0.26 (billed)
- Pause and resume
- not supported
Boat
Ahead of E2B on 7 of the 21 tests both finished, behind on 14, tied on 0.
- Cold start, median
- 1.1 s 6th of 9behind E2B
- Tool replay, median
- 43 s 9th of 9behind E2B
- Cost per 1,000 tasks
- $0.52 (estimate)
- Pause and resume
- Boat has no pause: stop and start restarts the machine and ends every running process.
E2B alternatives on each one's default machine
Each provider's default sandbox, so machine sizes differ. Runs from 2026-10-22 to 2026-10-23, 9 providers. Of the 8 alternatives, 0 start faster than E2B and 8 slower, none tied within the 95% intervals. Ranks are out of every provider on this track; see them all on the full results.
E2B on this track, for reference
- Cold start, median
- 0.50 s 1st of 9
- Tool replay, median
- 16 s 3rd of 9
- Cost per 1,000 tasks
- $0.49 (metered)
- Pause and resume
- pause 0.26 s, resume 0.45 s (median of 100)
A task is: create a sandbox, run the 100-call tool replay, delete it. Cost per task comes from a separate cost run, billed per sandbox. Cost kinds differ between providers (billed, metered, estimate) and aren't directly comparable.
Freestyle
Ahead of E2B on 7 of the 25 tests both finished, behind on 16, tied on 2.
- Cold start, median
- 0.72 s 4th of 9behind E2B
- Tool replay, median
- 26 s 6th of 9behind E2B
- Cost per 1,000 tasks
- $1.78 (estimate)
- Pause and resume
- pause 0.27 s, resume 0.32 s (median of 100)
Tenki
Ahead of E2B on 8 of the 25 tests both finished, behind on 15, tied on 2.
- Cold start, median
- 0.86 s 5th of 9behind E2B
- Tool replay, median
- 18 s 4th of 9behind E2B
- Cost per 1,000 tasks
- $0.86 (estimate)
- Pause and resume
- pause 55.56 s, resume 2.29 s (median of 100)
boxd
Ahead of E2B on 8 of the 25 tests both finished, behind on 17, tied on 0.
- Cold start, median
- 1.1 s 6th of 9behind E2B
- Tool replay, median
- 33 s 8th of 9behind E2B
- Cost per 1,000 tasks
- $1.03 (metered)
- Pause and resume
- pause 0.13 s, resume 0.41 s (median of 100)
Superserve
Ahead of E2B on 15 of the 25 tests both finished, behind on 9, tied on 1.
- Cold start, median
- 0.53 s 2nd of 9behind E2B
- Tool replay, median
- 13 s 1st of 9ahead of E2B
- Cost per 1,000 tasks
- $0.27 (billed)
- Pause and resume
- pause 0.36 s, resume 0.54 s (median of 100)
Upstash Box
Ahead of E2B on 9 of the 25 tests both finished, behind on 16, tied on 0.
- Cold start, median
- 2.6 s 9th of 9behind E2B
- Tool replay, median
- 27 s 7th of 9behind E2B
- Cost per 1,000 tasks
- $0.17 (billed)
- Pause and resume
- pause 0.73 s, resume 0.59 s (median of 100)
Daytona
Ahead of E2B on 11 of the 23 tests both finished, behind on 10, tied on 2.
- Cold start, median
- 1.5 s 8th of 9behind E2B
- Tool replay, median
- 15 s 2nd of 9ahead of E2B
- Cost per 1,000 tasks
- $0.64 (billed)
- Pause and resume
- Daytona's container sandboxes have no pause: stop and start restarts the sandbox and ends every running process.
Upstash Ephemeral Box
Ahead of E2B on 10 of the 24 tests both finished, behind on 13, tied on 1.
- Cold start, median
- 0.68 s 3rd of 9behind E2B
- Tool replay, median
- 23 s 5th of 9behind E2B
- Cost per 1,000 tasks
- $0.13 (billed)
- Pause and resume
- not supported
Boat
Ahead of E2B on 7 of the 22 tests both finished, behind on 15, tied on 0.
- Cold start, median
- 1.2 s 7th of 9behind E2B
- Tool replay, median
- 43 s 9th of 9behind E2B
- Cost per 1,000 tasks
- $0.49 (estimate)
- Pause and resume
- Boat has no pause: stop and start restarts the machine and ends every running process.
Quick picks, E2B included
The results page's own picks on the same 4 vCPU, 8 GB machine, by need. They follow the medians, so a close runner-up can be a tie.
| Need | Pick | Runner-up |
|---|---|---|
| Fastest cold start | Superserve 0.52 s | Upstash Ephemeral Box 0.68 s |
| Fastest agent work (tool replay) | Daytona 12 s | Superserve 13 s |
| Fastest 25-sandbox fan-out | Superserve 0.79 s | E2B 1.1 s |
| Fastest reconnect | Daytona 0.30 s | Superserve 0.33 s |
| Cheapest task | Upstash Ephemeral Box $0.00026 per task, $0.26 per 1,000 (billed) | Upstash Box $0.00034 per task, $0.34 per 1,000 (billed) |
| Most reliable | E2B, Tenki, Superserve, Upstash Box, Daytona, Upstash Ephemeral Box, Boat 100.00% (tied) | Freestyle 99.91% |
Providers we haven't tested
Sandbox providers we haven't run through the bench yet include Modal, Vercel Sandbox, Cloudflare Sandbox, Runloop, Northflank, Blaxel, Fly Sprites, CodeSandbox and Morph, so this page has no numbers for them.
How to read these numbers
- Every call starts from one server in Florida, US. Providers hosted farther away pay that round trip on every call; Boat (France) and boxd (Germany) pay a transatlantic one.
- There are two tracks. Same machine: 4 vCPU, 8 GB: Every provider asked for Boat's machine, 4 vCPU and 8 GB, or its nearest size. A provider marked “machine differs” measured something else inside. Default machines: Each provider's default sandbox, so machine sizes differ.
- Cost per task has a kind. Billed comes from the provider's bill, metered from its usage records at published rates, and estimate from our measured time at list price. Kinds are not directly comparable. Prices cover compute only at list rates: disk, transfer, subscriptions and credits are excluded.
- Pause and resume times come from the previous SDK run, because the newest runs skipped them. A burst size is published only when all three bursts finished.
- Failures caused by our own code are set aside, not counted. Every provider-side failure is kept.
- Ranks compare medians. Medians inside each other's 95% intervals count as tied, so a close rank can be noise.