P95 stands for the 95th percentile. 95% of iterations completed at or below this time. Tells you the 'typical worst case' — what a user experiences on a bad-but-not-extreme run.

P99 stands for the 99th percentile. 99% of iterations completed at or below this time. It highlights the tail latency (the rare outlier spikes). With 100 iterations, this is the 99th value (second-worst)

Burst TTI Benchmarks

Q: What is Time to Interactive (TTI)?

TTI is the total elapsed time it takes for a sandbox provider to boot up a sandbox and run a terminal command inside the sandbox.

Q: Why use median instead of average?

The average can be easily skewed by a single extreme outlier. The median provides a more accurate representation of the typical performance most users will experience.

Agents, sandboxes, & vibe-coded apps are powered by Cloud Run.

Burst TTI Benchmarks

A leaderboard of sandbox providers measured by time to interactive in a 100-sandbox burst.

Namespace · 4 vCPU / 16GB RAM · Northern Virginia, US

View on Github

Last run: August 28, 2026

Details

Performance Over Time

Composite Score

Detailed Metrics

Provider	Score	Median	P95	P99	Success
Isorun	99.3	0.06s	0.08s	0.08s	100%
CreateOS	97.8	0.21s	0.23s	0.23s	100%
Archil	97.0	0.26s	0.35s	0.37s	100%
OpenComputer	83.6	0.27s	0.29s	0.30s	86%
Arker	96.7	0.31s	0.35s	0.36s	100%
Daytona	87.5	0.33s	0.48s	0.49s	91%
Declaw	94.8	0.48s	0.57s	0.58s	100%
Superserve	93.7	0.49s	0.81s	0.87s	100%
Blaxel	94.5	0.52s	0.59s	0.63s	100%
Vercel	93.8	0.54s	0.70s	0.83s	100%
Cloud Run	91.0	0.68s	1.17s	1.29s	100%
Miosa	92.4	0.70s	0.85s	0.86s	100%
Modal	91.1	0.85s	0.95s	0.96s	100%
Mosaic	83.3	0.97s	1.48s	1.84s	95%
Runloop	89.1	1.06s	1.13s	1.13s	100%
E2B	88.2	1.09s	1.31s	1.37s	100%
Beam	84.5	1.34s	1.65s	1.65s	99%
Tensorlake	81.8	1.40s	2.43s	2.43s	100%
Tenki	81.3	1.67s	2.16s	2.17s	100%
Sail	72.0	2.68s	2.96s	2.99s	100%
Cloudflare	40.6	5.26s	6.71s	7.41s	100%
Upstash	36.5	5.27s	7.94s	8.01s	100%
CodeSandbox	18.2	6.96s	10.04s	10.64s	100%
Run Cloud	0.0	11.81s	22.60s	26.35s	100%
Sandbox0	0.0	27.38s	27.49s	27.52s	24%
Hopx	0.0	0.00s	0.00s	0.00s	0%
Lightning AI	0.0	0.00s	0.00s	0.00s	0%
Microsandbox	0.0	0.00s	0.00s	0.00s	0%
Northflank	0.0	0.00s	0.00s	0.00s	0%

Want to see a provider added?

Let us know on X

Methodology

What We Measure

Every benchmark measures Time to Interactive (TTI) — the elapsed time from calling compute.sandbox.create() to the first successful runCommand() inside the sandbox.

Each provider is tested with 100 iterations per run. Benchmarks run automatically via GitHub Actions on a recurring schedule. All results are committed to the public benchmarks repo.

Burst Test: All sandboxes are launched concurrently in a single burst.

How We Score

The Composite Score is a weighted blend of timing metrics multiplied by the success rate. Each metric is scored against a fixed 10-second ceiling: 100 × (1 − value / 10,000ms), so a 200ms median scores 98 and anything ≥10s scores 0.

The weighted timing score is then multiplied by the success rate (0–1), so providers that fail frequently are penalized proportionally.