Observed arrival · 2026-10-24
Sandbox Bench: Cloud sandboxes, measured
A benchmark compares nine cloud sandbox providers on startup speed, agent tasks, reliability, and cost.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Developers choosing sandboxes for AI agents
- Worth noticing
- boxd has no 4-vCPU, 8-GB option, so its shared-machine result uses the nearest size: 4 vCPU and 16 GB.
Field notes
The page separates results on a shared 4-vCPU, 8-GB machine from results on each provider’s default machine. Its sections cover start-up, agent work, pause and snapshot behavior, bursts, machines, reliability, and cost. The reported test window is Oct. 22–23, 2026, with 100 runs per test and 30 per agent task in Florida, US. One hardware mismatch is disclosed: boxd used 4 vCPU and 16 GB because it has no 8-GB option at that CPU size.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue