Providers / Vercel

Vercel

Frontend cloud; measured here via Sandbox, its ephemeral microVM compute.

vercel.com10 fresh-host observations

CPU all-core vs. monthly price

1 / 1 plans

Cross-host P50 against the net monthly price (EUR, excluding VAT — tax depends on the buyer, not the provider). Ringed dots sit on the efficient frontier: no cheaper measured plan is faster on this metric. Points are deliberately never connected.

View

Ringed dots are on the efficient frontier for this metric.Faded points need 10 independent fresh hosts before ranking.

Performance history

Every campaign is a separate point in time. Old hardware is retained, never averaged into a newer campaign, so generation changes remain visible.

2 dated cohorts

Composite over time

P50 per campaign · index

2 vCPU / 4 GB · iad1

The first campaign is the baseline. A trend appears after the same plan and region are measured again under a new campaign ID.

MeasuredPlan / regionHardware observedEvidence

24 Jul 2026 UTC2 vCPU / 4 GB · iad1Intel(R) Xeon(R) Processor @ 2.90GHz10 hosts

24 Jul 2026 UTC2 vCPU / 4 GB · iad1Intel(R) Xeon(R) Processor @ 2.50GHz10 hosts

2 vCPU / 4 GB

usage-basedSandboxiad1measured 24 Jul 2026 UTC€218.38/mo · €0.2992/hranked · 10 hostsCompare in Compute →

Performance

100.0

Value index

4.0

Independent hosts

Raw runs

CPU: Intel(R) Xeon(R) Processor @ 2.90GHz
Cores: 2
Memory: 4 GiB
OS / arch: linux/amd64
Kernel: 6.12.76
CLI: 0.2.0-dev
host_strategy: fresh-vercel-sandbox-per-sample
pricing_as_of: 2026-07-23
pricing_basis: 2*$0.128 active-vCPU-hour + 4GB*$0.0212 GB-hour; ECB 2026-07-23 EUR/USD 1.1392; 730h continuous full utilization; excludes included quota, creation, and network
runtime: node24
sandbox_memory_gb: 4
sandbox_vcpus: 2
vercel_cli_version: 56.4.1

cpu

measured 24 Jul 2026 UTC

Metric	P50	Mean	P10	P90	P99	vs best
multi_core_hash	2,637MB/s	1,874	708	2,658	2,660	−26%
scaling_efficiency	101%	101	99.9	101	101	−1%
single_core_hash	1,314MB/s	933	352	1,329	1,329	−27%

SHA-256 over 64 KiB buffers; uses CPU crypto extensions where available

disk

measured 24 Jul 2026 UTC

Metric	P50	Mean	P10	P90	P99	vs best
fsync_latency_p50	0.04ms	0.04	0.03	0.05	0.06	best
rand_read_4k	69.8KIOPS	62.6K	47.4K	75.1K	77.3K	best
seq_read	3,479MB/s	3,633	2,339	5,595	5,914	best
seq_write	1,607MB/s	1,514	1,312	1,674	1,707	best

reads bypass the page cache (direct I/O) · test file: 1024 MiB, random reads at queue depth 4

memory

measured 24 Jul 2026 UTC

Metric	P50	Mean	P10	P90	P99	vs best
copy_bandwidth	6.66GB/s	5.48	3.55	6.85	6.89	−73%
random_access_latency	108ns	110	106	115	119	−3%

bandwidth counts bytes copied (read+write traffic is ~2x)

network

measured 24 Jul 2026 UTC

Metric	P50	Mean	P10	P90	P99	vs best
download	1,882Mbps	1,946	1,652	2,260	2,704	−13%
latency_p50	4.70ms	4.59	4.27	4.95	4.96	best
upload	253Mbps	291	218	378	578	−39%

latency excludes server processing time (Cloudflare Server-Timing) · measured against Cloudflare edge IAD

steal

measured 24 Jul 2026 UTC

Metric	P50	Mean	P10	P90	P99	vs best
consistency_cv	3.03%	3.44	2.11	5.33	6.42	best
cpu_steal	0.60%	0.53	0.39	0.61	0.69	−0%
p99_over_p50	1.09ratio	1.13	1.07	1.29	1.30	−8%
unit_time_p50	18.9ms	18.3	16.1	19.5	19.5	−152%

consistency_cv near 0% and p99_over_p50 near 1.0 indicate a quiet host; high values suggest noisy neighbors or throttling