Measurements
What our VPS actually do
All four plans put through the bench, each reading with its method. The figures that do not flatter us are here too: a chosen measurement is no longer a measurement.
Method · Every reading was taken on production VPS identical to the ones we sell, ordered through the public checkout, with no special tuning. Each table states the tool and its parameters. Readings taken on 12 and 13 November 2026.
Advertised against measured
The first table is the one that matters most: it puts what the price list promises against what the machine delivers, plan by plan.
What the storefront promises, checked on every plan
Taken on four VPS ordered separately, at delivery.
| Plan | vCPU and memory | Disk | Throughput | Geekbench 6 |
|---|---|---|---|---|
| Start · 2 vCPU | 2 vCPU · 4 GB | 40 GB real / 40 advertised | 959 / 1,000 Mbit/s | 2,272 measured |
| Medium · 4 vCPU | 4 vCPU · 8 GB | 80 GB real / 80 advertised | 961 / 1,000 Mbit/s | 3,879 measured |
| Pro · 6 vCPU | 6 vCPU · 16 GB | 120 GB real / 120 advertised | 963 / 1,000 Mbit/s | 4,789 measured |
| Ultra · 8 vCPU | 8 vCPU · 32 GB | 160 GB real / 160 advertised | 959 / 1,000 Mbit/s | 5,618 measured |
Processor
Two Intel Xeon Gold 6238 per hypervisor. The scores below were taken on our own VPS and published: every row links to its result, which anyone can open.
Geekbench 6
Geekbench 6.5, one run per plan, results published on browser.geekbench.com.
| Plan | Single-core | Multi-core | Public result |
|---|---|---|---|
| Start · 2 vCPU | 1,275 | 2,272 | 19461712 ↗ |
| Medium · 4 vCPU | 1,259 | 3,879 | 19462046 ↗ |
| Pro · 6 vCPU | 1,250 | 4,789 | 19462136 ↗ |
| Ultra · 8 vCPU | 1,232 | 5,618 | 19462240 ↗ |
Sustained load and neighbours
sysbench, 100% of the vCPU for 5 minutes, then all four VPS on one hypervisor started together.
| Measurement | Result |
|---|---|
| Throughput held after 5 minutes at full load | 97% |
| Throughput kept with all four VPS flat out | 89 to 99% |
| Steal time taken by the hypervisor | 0.02% |
Disk
Four 12 Gb/s SAS SSDs in RAID 1+0 behind an HPE Smart Array controller. Direct I/O bypasses the cache: without it, these figures would be measuring memory.
4K random and sequential
fio, direct I/O, 128 parallel requests. 1 MB blocks for the sequential run.
| Plan | 4K read | 4K write | Seq. read | Seq. write |
|---|---|---|---|---|
| Start · 2 vCPU | 172,701 IOPS | 165,714 IOPS | 3,787 MB/s | 2,161 MB/s |
| Medium · 4 vCPU | 163,205 IOPS | 164,824 IOPS | 3,783 MB/s | 2,156 MB/s |
| Pro · 6 vCPU | 162,073 IOPS | 158,873 IOPS | 3,758 MB/s | 2,161 MB/s |
| Ultra · 8 vCPU | 164,409 IOPS | 163,297 IOPS | 3,778 MB/s | 2,159 MB/s |
Latency under mixed load
Mixed load, 70% read / 30% write, 64 parallel requests.
| Plan | Median latency | p99 latency | Mixed 70/30 |
|---|---|---|---|
| Start · 2 vCPU | 305 µs | 408 µs | 142,994 IOPS |
| Medium · 4 vCPU | 354 µs | 444 µs | 140,627 IOPS |
| Pro · 6 vCPU | 354 µs | 424 µs | 138,140 IOPS |
| Ultra · 8 vCPU | 375 µs | 449 µs | 140,670 IOPS |
Steadiness over 24 hours
Taken every hour for 24 hours on the Start plan, peak hours included.
| Measurement | Median | Peak 18:00-23:00 | Night 02:00-06:00 |
|---|---|---|---|
| Processor (sysbench, events/s) | 1,932 | 1,937 | 1,944 |
| Disk 4K (IOPS) | 149,110 | 147,122 | 150,122 |
| 4K latency | 339 µs | 341 µs | 344 µs |
Network
The link is advertised at a guaranteed 1 Gbit/s, with access to the hypervisor's 10 GbE above that. For a French audience the question that matters is not throughput, it is latency from a home connection.
From consumer home connections
Globalping: 17 home connections across France, 16 pings an hour for 24 hours. The reference column measures the same connections against a third-party server in Paris, which removes the subscriber's own line and leaves only the path to us.
| Operator | Lines | To the VPS | To the reference | Difference |
|---|---|---|---|---|
| Bouygues Telecom | 3 | 3.9 ms | 3.3 ms | +0.8 ms |
| Free | 5 | 4.4 ms | 3.1 ms | +1.4 ms |
| SFR | 4 | 9.8 ms | 10.7 ms | −0.3 ms |
| Orange | 5 | 12.2 ms | 10.9 ms | +1.3 ms |
Round trip from the VPS, over 24 hours
Median over 24 consecutive hours, peak hours included.
| Destination | Median | p99 | p99.9 |
|---|---|---|---|
| Paris | 1.2 ms | 1.4 ms | 1.7 ms |
| London | 8.2 ms | 9.0 ms | 12.3 ms |
| Amsterdam | 9.9 ms | 10.3 ms | 10.5 ms |
| Frankfurt | 10.7 ms | 11.1 ms | 11.3 ms |
| New York | 78.9 ms | 79.6 ms | 83.7 ms |
International throughput
iperf3 against the public servers used by the YABS script, inbound then outbound.
| Destination | Inbound | Outbound |
|---|---|---|
| London | 953 Mbit/s | 1,120 Mbit/s |
| Amsterdam | 937 Mbit/s | 1,120 Mbit/s |
| New York | 824 Mbit/s | 1,050 Mbit/s |
| Singapore | 720 Mbit/s | 885 Mbit/s |
By use case
The same machines put to work on what people actually run. Each row compares the entry plan with the largest one.
From Start to Ultra
Each use case with its real software and real load: WordPress with WooCommerce and no page cache, PostgreSQL 17 with 16 clients, llama.cpp with no graphics card, Paper 26.3 for Minecraft, WireGuard between two VPS, the official Nextcloud 35 image.
| Use case | Measurement | Start | Ultra |
|---|---|---|---|
| WooCommerce shop, no cache | dynamic pages/s | 21 | 107 |
| PostgreSQL 17, 16 clients | transactions/s | 3,694 | 8,611 |
| PostgreSQL 17, read-only | transactions/s | 24,624 | 100,469 |
| MariaDB 11.8, 16 connections | transactions/s | 1,205 | 3,723 |
| C++ build (llama.cpp) | duration | 555 s | 183 s |
| Next.js build | duration | 11 s | 11 s |
| WireGuard tunnel | throughput | 935 Mbit/s | 935 Mbit/s |
| Nextcloud, 1 GB file | download | 84 MB/s | 85 MB/s |
| Minecraft Paper, startup | duration | 17.1 s | 14.8 s |
| Qwen3 4B, no graphics card | tokens/s | 4.2 | 12.1 |
| 720p live transcoding | simultaneous streams | 2.6 | 6.8 |
| Virtual machine inside the VPS | processor kept | 97% | 97% |