AI Agent infrastructure, measured.
Independent benchmarks and teardowns of the infrastructure agents run on: sandboxes, agent VMs, computer-use and browser infra, and agent evals. Every number links to the raw run.
See the leaderboardWork with me
Time to first command
How long from asking for a fresh sandbox until a shell command actually runs in it.
Cold create to first command
ms · lower is better · p50 of 5 fresh sandboxes each
- 1Cloudflare Sandbox483 ms×1.00
- 2Vercel Sandbox662 ms×1.37
- 3Freestyle976 ms×2.02
- 4Tensorlake1.92 s×3.97
- 5BoxLite Cloud4.07 s×8.42
- 6Upstash Box5.02 s×10.39
- 1Cloudflare Sandbox467 ms×1.00
- 2Vercel Sandbox632 ms×1.35
- 3Freestyle897 ms×1.92
- 4Tensorlake1.90 s×4.06
- 5Upstash Box2.76 s×5.90
- 6BoxLite Cloud3.75 s×8.03
- 1Cloudflare Sandbox888 ms×1.00
- 2Vercel Sandbox1.07 s×1.20
- 3Freestyle1.57 s×1.77
- 4Tensorlake2.49 s×2.80
- 5BoxLite Cloud4.43 s×4.98
- 6Upstash Box11.67 s×13.13
Method, raw results and the providers we could not measure →