Tenki sandbox benchmark
Our hardware probe found Tenki runs each sandbox as a Cloud Hypervisor microVM. We timed it through one SDK against 8 other sandbox providers: cold start, real agent work, everyday calls, pause and snapshots, bursts, machine speed, reliability and cost per task. Each result is a median with its p95 and 95% interval, on two tracks: the same 4 vCPU, 8 GB machine for every provider, and each provider's default machine.
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.
Same machine: 4 vCPU, 8 GB
- Cold start, median
- 1.1 s 5th of 9, tied
- Tool replay, median
- 21 s 4th of 9
Default machines
- Cold start, median
- 0.86 s 5th of 9
- Tool replay, median
- 18 s 4th of 9
Tenki 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. See every provider side by side on the full results.
- Asked for
- 4 vCPU, 8 GB
- Measured inside
- 4 vCPU, 7.8 GB
- Cost per 1,000 tasks
- $1.79 (estimate)
- Success rate
- 100.00% (1,053 of 1,053)
- Cost: $0.0018 per task, our measured time at list price (estimate over 10 tasks, low confidence). 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.
- List price: $0.331 per hour for this machine, compute only.
- Pause and resume: supported; pause 60.53 s and resume 2.45 s median over 77 runs.
- Failures counted against Tenki: none.
Start-up
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Cold start (lower is better) | 1.1 sp95 3.3 s95%: 0.94 s to 1.2 s | 3.3 s | 0.94 s to 1.2 s | 5th of 9, tied with Boat · lowest median: Superserve 0.52 s |
Agent work
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Agent tool replay (lower is better) | 21 sp95 31 s95%: 19 s to 22 s | 31 s | 19 s to 22 s | 4th of 9 · lowest median: Daytona 12 s |
| npm install (lower is better) | 1.1 sp95 1.8 s95%: 1.0 s to 1.4 s | 1.8 s | 1.0 s to 1.4 s | 6th of 9, tied with Superserve · lowest median: Daytona 0.75 s |
| pip install (lower is better) | 3.3 sp95 5.0 s95%: 3.1 s to 4.5 s | 5.0 s | 3.1 s to 4.5 s | 1st of 9 |
| git clone (lower is better) | 1.2 sp95 2.1 s95%: 1.1 s to 1.3 s | 2.1 s | 1.1 s to 1.3 s | 5th of 9, tied with E2B · lowest median: Daytona 0.43 s |
| Dev server ready (lower is better) | 1.9 sp95 4.5 s95%: 1.9 s to 2.0 s | 4.5 s | 1.9 s to 2.0 s | 2nd of 9 · lowest median: Superserve 1.3 s |
| Preview URL first byte (lower is better) | 0.38 sp95 0.49 s95%: 0.37 s to 0.39 s | 0.49 s | 0.37 s to 0.39 s | 6th of 9 · lowest median: Superserve 0.18 s |
| Edit, rebuild, reload (lower is better) | 0.52 sp95 0.61 s95%: 0.50 s to 0.54 s | 0.61 s | 0.50 s to 0.54 s | 7th of 9 · lowest median: Freestyle 0.30 s |
Everyday calls
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Run a command (lower is better) | 0.18 sp95 0.21 s95%: 0.18 s to 0.19 s | 0.21 s | 0.18 s to 0.19 s | 4th of 9 · lowest median: E2B 0.13 s |
| Write + read 1 KB (lower is better) | 0.27 sp95 0.34 s95%: 0.22 s to 0.29 s | 0.34 s | 0.22 s to 0.29 s | 5th of 9 · lowest median: Superserve 0.10 s |
| Write + read 10 MB (lower is better) | 7.7 sp95 13 s95%: 6.7 s to 8.5 s | 13 s | 6.7 s to 8.5 s | 9th of 9, tied with Boat · lowest median: Superserve 0.66 s |
| Reconnect (lower is better) | 0.64 sp95 0.73 s95%: 0.63 s to 0.65 s | 0.73 s | 0.63 s to 0.65 s | 7th of 9 · lowest median: Daytona 0.30 s |
| Destroy (lower is better) | 0.12 sp95 0.16 s95%: 0.11 s to 0.13 s | 0.16 s | 0.11 s to 0.13 s | 2nd of 9 · lowest median: E2B 0.11 s |
Pause and snapshots
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Pause + resume (lower is better) | 64 sp95 84 s95%: 61 s to 68 s | 84 s | 61 s to 68 s | 6th of 6 · lowest median: boxd 0.53 s |
| Take snapshot (lower is better) | 6.9 sp95 8.8 s95%: 6.7 s to 6.9 s | 8.8 s | 6.7 s to 6.9 s | 7th of 8 · lowest median: Superserve 0.39 s |
| Start from snapshot (lower is better) | 1.9 sp95 2.9 s95%: 1.7 s to 2.0 s | 2.9 s | 1.7 s to 2.0 s | 6th of 8 · lowest median: Superserve 0.55 s |
Bursts
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| 10 at once, all ready (lower is better) | 1.2 sp95 1.2 s95%: 1.1 s to 1.2 s | 1.2 s | 1.1 s to 1.2 s | 4th of 8, tied with Freestyle · lowest median: Superserve 0.84 s |
| 25 at once, all ready (lower is better) | 1.2 sp95 1.3 s95%: 1.2 s to 1.3 s | 1.3 s | 1.2 s to 1.3 s | 4th of 9 · lowest median: Superserve 0.79 s |
Machine speed
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| CPU, one core (higher is better) | 19.4M/sp95 20.9M/s95%: 17.2M/s to 19.8M/s | 20.9M/s | 17.2M/s to 19.8M/s | 1st of 9, tied with Boat |
| CPU, all cores (higher is better) | 36.6M/sp95 50.3M/s95%: 34.5M/s to 39.5M/s | 50.3M/s | 34.5M/s to 39.5M/s | 5th of 9 · highest median: Upstash Ephemeral Box 51M/s |
| Memory bandwidth (higher is better) | 45.9 GB/sp95 51.7 GB/s95%: 39.1 GB/s to 47.9 GB/s | 51.7 GB/s | 39.1 GB/s to 47.9 GB/s | 1st of 9 |
| Disk write (higher is better) | 444 MB/sp95 599 MB/s95%: 407 MB/s to 461 MB/s | 599 MB/s | 407 MB/s to 461 MB/s | 6th of 9, tied with Boat · highest median: Upstash Ephemeral Box 2.4 GB/s |
| Disk read (higher is better) | 12.1 GB/sp95 20.5 GB/s95%: 10.5 GB/s to 14.3 GB/s | 20.5 GB/s | 10.5 GB/s to 14.3 GB/s | 5th of 9 · highest median: Upstash Ephemeral Box 23.1 GB/s |
| 4K random write (higher is better) | 370.6 IOPSp95 443.7 IOPS95%: 357.5 IOPS to 376.5 IOPS | 443.7 IOPS | 357.5 IOPS to 376.5 IOPS | 8th of 9 · highest median: Upstash Ephemeral Box 45.1K IOPS |
| 4K random read (higher is better) | 221K IOPSp95 617K IOPS95%: 217.5K IOPS to 226.7K IOPS | 617K IOPS | 217.5K IOPS to 226.7K IOPS | 9th of 9 · highest median: Boat 796.4K IOPS |
Quick picks on this track, all providers
| 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% |
Tenki on its default machine
Each provider's default sandbox, so machine sizes differ. Runs from 2026-10-22 to 2026-10-23, 9 providers. See every provider side by side on the full results.
- Asked for
- provider default
- Measured inside
- 2 vCPU, 3.8 GB
- Cost per 1,000 tasks
- $0.86 (estimate)
- Success rate
- 100.00% (1,076 of 1,076)
- Machine: Tenki documents a 5 GiB default disk but not a default vCPU or RAM, so those are what we measured.
- Cost: $0.00086 per task, our measured time at list price (estimate over 10 tasks, low confidence). 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.
- List price: $0.162 per hour for this machine, compute only.
- Pause and resume: supported; pause 55.56 s and resume 2.29 s median over 100 runs.
- Failures counted against Tenki: [internal] internal server error (2 tests)
Start-up
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Cold start (lower is better) | 0.86 sp95 3.1 s95%: 0.81 s to 0.92 s | 3.1 s | 0.81 s to 0.92 s | 5th of 9 · lowest median: E2B 0.50 s |
Agent work
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Agent tool replay (lower is better) | 18 sp95 24 s95%: 17 s to 20 s | 24 s | 17 s to 20 s | 4th of 9 · lowest median: Superserve 13 s |
| npm install (lower is better) | 1.4 sp95 2.1 s95%: 1.0 s to 1.5 s | 2.1 s | 1.0 s to 1.5 s | 5th of 9, tied with Superserve, boxd · lowest median: Boat 0.86 s |
| pip install (lower is better) | 4.3 sp95 4.9 s95%: 3.1 s to 4.5 s | 4.9 s | 3.1 s to 4.5 s | 4th of 9, tied with Boat, Upstash Box, Upstash Ephemeral Box · lowest median: Superserve 3.8 s |
| git clone (lower is better) | 1.2 sp95 2.3 s95%: 1.0 s to 1.3 s | 2.3 s | 1.0 s to 1.3 s | 6th of 9, tied with Superserve, E2B · lowest median: Upstash Ephemeral Box 0.61 s |
| Dev server ready (lower is better) | 1.9 sp95 3.2 s95%: 1.9 s to 2.0 s | 3.2 s | 1.9 s to 2.0 s | 2nd of 9 · lowest median: Superserve 1.8 s |
| Preview URL first byte (lower is better) | 0.40 sp95 1.5 s95%: 0.39 s to 0.44 s | 1.5 s | 0.39 s to 0.44 s | 6th of 9 · lowest median: Superserve 0.19 s |
| Edit, rebuild, reload (lower is better) | 0.52 sp95 1.7 s95%: 0.51 s to 0.53 s | 1.7 s | 0.51 s to 0.53 s | 7th of 9 · lowest median: Freestyle 0.31 s |
Everyday calls
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Run a command (lower is better) | 0.17 sp95 0.20 s95%: 0.16 s to 0.17 s | 0.20 s | 0.16 s to 0.17 s | 4th of 9 · lowest median: E2B 0.14 s |
| Write + read 1 KB (lower is better) | 0.21 sp95 0.32 s95%: 0.21 s to 0.21 s | 0.32 s | 0.21 s to 0.21 s | 4th of 9, tied with Freestyle · lowest median: Superserve 0.10 s |
| Write + read 10 MB (lower is better) | 3.6 sp95 9.2 s95%: 1.9 s to 4.6 s | 9.2 s | 1.9 s to 4.6 s | 8th of 9 · lowest median: Superserve 0.64 s |
| Reconnect (lower is better) | 0.67 sp95 2.9 s95%: 0.66 s to 0.68 s | 2.9 s | 0.66 s to 0.68 s | 7th of 9 · lowest median: Daytona 0.31 s |
| Destroy (lower is better) | 0.14 sp95 0.21 s95%: 0.13 s to 0.14 s | 0.21 s | 0.13 s to 0.14 s | 2nd of 9 · lowest median: E2B 0.11 s |
Pause and snapshots
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| Pause + resume (lower is better) | 58 sp95 82 s95%: 56 s to 59 s | 82 s | 56 s to 59 s | 6th of 6 · lowest median: boxd 0.54 s |
| Take snapshot (lower is better) | 6.7 sp95 7.0 s95%: 6.7 s to 6.8 s | 7.0 s | 6.7 s to 6.8 s | 7th of 8 · lowest median: E2B 0.34 s |
| Start from snapshot (lower is better) | 1.8 sp95 2.5 s95%: 1.7 s to 2.1 s | 2.5 s | 1.7 s to 2.1 s | 6th of 8 · lowest median: Superserve 0.56 s |
Bursts
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| 10 at once, all ready (lower is better) | 1.0 sp95 1.5 s95%: 1.0 s to 1.5 s | 1.5 s | 1.0 s to 1.5 s | 3rd of 9, tied with E2B · lowest median: Superserve 0.65 s |
| 25 at once, all ready (lower is better) | 1.3 sp95 1.9 s95%: 0.98 s to 1.9 s | 1.9 s | 0.98 s to 1.9 s | 4th of 8 · lowest median: Superserve 0.75 s |
Machine speed
| Test | Median | p95 | 95% interval | Rank |
|---|---|---|---|---|
| CPU, one core (higher is better) | 18.6M/sp95 21.4M/s95%: 11.3M/s to 19.9M/s | 21.4M/s | 11.3M/s to 19.9M/s | 2nd of 9, tied with Boat · highest median: Boat 18.8M/s |
| CPU, all cores (higher is better) | 20.9M/sp95 31.8M/s95%: 18.1M/s to 22.1M/s | 31.8M/s | 18.1M/s to 22.1M/s | 5th of 9 · highest median: Boat 51.7M/s |
| Memory bandwidth (higher is better) | 38.9 GB/sp95 52.1 GB/s95%: 24.3 GB/s to 48.0 GB/s | 52.1 GB/s | 24.3 GB/s to 48.0 GB/s | 2nd of 9 · highest median: Upstash Ephemeral Box 41.1 GB/s |
| Disk write (higher is better) | 415 MB/sp95 578 MB/s95%: 267 MB/s to 457 MB/s | 578 MB/s | 267 MB/s to 457 MB/s | 7th of 9 · highest median: Upstash Ephemeral Box 2.4 GB/s |
| Disk read (higher is better) | 10.6 GB/sp95 21.6 GB/s95%: 7.8 GB/s to 11.7 GB/s | 21.6 GB/s | 7.8 GB/s to 11.7 GB/s | 6th of 9, tied with boxd · highest median: Upstash Ephemeral Box 23.0 GB/s |
| 4K random write (higher is better) | 361.3 IOPSp95 433.3 IOPS95%: 332.6 IOPS to 379.8 IOPS | 433.3 IOPS | 332.6 IOPS to 379.8 IOPS | 9th of 9 · highest median: Upstash Ephemeral Box 43.3K IOPS |
| 4K random read (higher is better) | 223.5K IOPSp95 623.4K IOPS95%: 221.5K IOPS to 226.6K IOPS | 623.4K IOPS | 221.5K IOPS to 226.6K IOPS | 9th of 9 · highest median: Boat 772.6K IOPS |
Quick picks on this track, all providers
| Need | Pick | Runner-up |
|---|---|---|
| Fastest cold start | E2B 0.50 s | Superserve 0.53 s |
| Fastest agent work (tool replay) | Superserve 13 s | Daytona 15 s |
| Fastest 25-sandbox fan-out | Superserve 0.75 s | E2B 0.87 s |
| Fastest reconnect | Daytona 0.31 s | Superserve 0.32 s |
| Cheapest task | Upstash Ephemeral Box $0.00013 per task, $0.13 per 1,000 (billed) | Upstash Box $0.00017 per task, $0.17 per 1,000 (billed) |
| Most reliable | E2B, Tenki, Superserve, Upstash Box, Daytona, Boat 100.00% (tied) | boxd 99.81% |
How to read these numbers
- Freestyle, Tenki and Boat costs are list-price estimates on our measured sandbox lifetime, because none gives a per-sandbox bill (Boat drops it when a sandbox is deleted).
- 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.
More results
- E2B sandbox benchmark
- Daytona sandbox benchmark
- Superserve sandbox benchmark
- Upstash Box sandbox benchmark
- Upstash Ephemeral Box sandbox benchmark
- Boat sandbox benchmark
- Freestyle sandbox benchmark
- Every sandbox provider we benchmark
- Every provider on the same 4 vCPU, 8 GB machine
- Every provider on its default machine
- Raw data: results.json · results.md · llms.txt