This page collects every measurement the Lab has published so far: five Hetzner Cloud servers in Helsinki (CX23, CX33, CX43 on x86 and CAX11, CAX21 on ARM), all rented and tested on 5 November 2026 with the same scripts, plus our production n8n server. Every table on the rest of the site is rendered from the same data file, so a number reads the same everywhere. The method is described in how we test.

System: CPU, memory, disk, network

ServerCPU 1 thread
events/s
CPU all threads
events/s
Memory
MiB/s
Disk 4k read
IOPS
Disk 4k write
IOPS
Disk seq. readDownloadCPU stealPrice / mo
Hetzner CX232 vCPU · 4 GB · x86_649141,8234,09853,05047,9962,965 MB/s285 Mbit/s0.0 %6.64 €incl. VAT
Hetzner CX334 vCPU · 8 GB · x86_641,6216,4534,97441,64435,0131,936 MB/s781 Mbit/s0.0 %10.27 €incl. VAT
Hetzner CX438 vCPU · 15 GB · x86_643,47026,6974,96845,00542,1042,338 MB/s835 Mbit/s0.0 %19.35 €incl. VAT
Hetzner CAX112 vCPU · 4 GB · aarch643,2486,5504,17148,55442,0512,370 MB/s567 Mbit/s0.0 %7.25 €incl. VAT
Hetzner CAX214 vCPU · 8 GB · aarch643,30713,2104,26453,79646,0192,490 MB/s381 Mbit/s0.0 %12.69 €incl. VAT

Measured by AI Hosting Lab on 2026-11-05 · 5 servers · method · raw data

CX23913.9CX331,620.9CX433,470.4CAX113,248.3CAX213,307.3
cpu 1t events s: · x86 ARM
CX231,823.0CX336,453.1CX4326,697.4CAX116,549.8CAX2113,210.5
cpu nt events s: · x86 ARM

Ollama on CPU: tokens per second

ServerModelGeneration
tokens/s
Prompt
tokens/s
Load timeRAM used (system)Note
Hetzner CX232 vCPU · 4 GB · x86_64Llama 3.2 1B10.3454.03.5 s2,001 MiB
Hetzner CX232 vCPU · 4 GB · x86_64Llama 3.2 3B5.8325.28.8 s3,002 MiB
Hetzner CX232 vCPU · 4 GB · x86_64Llama 3.1 8Bdid not run–––OOM-killed after ~50 s: Ollama fit check reported it would leave 1698 MiB free,
Hetzner CX334 vCPU · 8 GB · x86_64Llama 3.2 1B23.21,322.82.0 s2,205 MiB
Hetzner CX334 vCPU · 8 GB · x86_64Llama 3.2 3B13.5776.53.7 s3,068 MiB
Hetzner CX334 vCPU · 8 GB · x86_64Llama 3.1 8B6.3258.610.0 s5,941 MiB
Hetzner CX438 vCPU · 15 GB · x86_64Llama 3.2 1B37.51,910.91.7 s2,293 MiB
Hetzner CX438 vCPU · 15 GB · x86_64Llama 3.2 3B21.21,161.33.7 s3,196 MiB
Hetzner CX438 vCPU · 15 GB · x86_64Llama 3.1 8B9.7447.47.9 s6,023 MiB
Hetzner CAX112 vCPU · 4 GB · aarch64Llama 3.2 1B15.0835.54.8 s2,203 MiB
Hetzner CAX112 vCPU · 4 GB · aarch64Llama 3.2 3B6.5376.46.0 s3,214 MiB
Hetzner CAX112 vCPU · 4 GB · aarch64Llama 3.1 8Bdid not run–––llama-server process has terminated: exit status 1: ggml_aligned_malloc: insuffi
Hetzner CAX214 vCPU · 8 GB · aarch64Llama 3.2 1B26.81,427.13.2 s2,315 MiB
Hetzner CAX214 vCPU · 8 GB · aarch64Llama 3.2 3B12.2664.15.2 s3,282 MiB
Hetzner CAX214 vCPU · 8 GB · aarch64Llama 3.1 8B5.7226.512.3 s5,796 MiB

Measured by AI Hosting Lab on 2026-11-05 · 5 servers · method · raw data

CX235.8CX3313.5CX4321.2CAX116.5CAX2112.2
Llama 3.2 3B: tokens/s (higher is better) · x86 ARM

n8n: webhook load test

ServerConcurrent requestsExecutions/sMedianp95Failedn8n peak RAM
Hetzner CX232 vCPU · 4 GB · x86_64110.1104 ms162 ms0520 MiB
Hetzner CX232 vCPU · 4 GB · x86_641014.8697 ms893 ms0660 MiB
Hetzner CX232 vCPU · 4 GB · x86_642515.01,651 ms1,936 ms0825 MiB
Hetzner CX232 vCPU · 4 GB · x86_645014.23,445 ms3,954 ms01,003 MiB
Hetzner CX334 vCPU · 8 GB · x86_64113.572 ms133 ms0543 MiB
Hetzner CX334 vCPU · 8 GB · x86_641031.9311 ms438 ms0574 MiB
Hetzner CX334 vCPU · 8 GB · x86_642529.5841 ms986 ms0803 MiB
Hetzner CX334 vCPU · 8 GB · x86_645030.11,647 ms1,846 ms01,050 MiB
Hetzner CX438 vCPU · 15 GB · x86_64114.678 ms123 ms0546 MiB
Hetzner CX438 vCPU · 15 GB · x86_641036.1280 ms341 ms0544 MiB
Hetzner CX438 vCPU · 15 GB · x86_642535.0714 ms800 ms0694 MiB
Hetzner CX438 vCPU · 15 GB · x86_645032.41,539 ms1,682 ms01,016 MiB
Hetzner CAX112 vCPU · 4 GB · aarch64110.3113 ms156 ms0585 MiB
Hetzner CAX112 vCPU · 4 GB · aarch641013.9755 ms934 ms0640 MiB
Hetzner CAX112 vCPU · 4 GB · aarch642512.91,909 ms2,279 ms0729 MiB
Hetzner CAX112 vCPU · 4 GB · aarch645013.23,786 ms4,114 ms01,038 MiB
Hetzner CAX214 vCPU · 8 GB · aarch64112.0107 ms128 ms0539 MiB
Hetzner CAX214 vCPU · 8 GB · aarch641025.4402 ms528 ms0579 MiB
Hetzner CAX214 vCPU · 8 GB · aarch642525.8965 ms1,106 ms0824 MiB
Hetzner CAX214 vCPU · 8 GB · aarch645024.52,019 ms2,197 ms0859 MiB

Measured by AI Hosting Lab on 2026-11-05 · 5 servers · method · raw data

n8n: production install

n8n 2.46.6 on Hetzner CX23Measured
Docker install22 s
Image pull + first start to healthy105 s
n8n RAM right after start785.3 MiB
n8n RAM idle (after ~60 s)379.7 MiB
PostgreSQL 16 RAM idle69.6 MiB
Whole server RAM used, idle1036 MiB of 3,814
n8n image size1.94GB
Disk used after install4.7G of 38G

Measured by AI Hosting Lab on 2026-11-05 · docker compose: n8n + postgres:16, docker.io from Ubuntu repo, task runners on · method

Supabase: official docker compose, idle

ServerContainersRAM (containers)RAM used (system)Fits?Image pullDisk (images)
Hetzner CX232 vCPU · 4 GB · x86_64111,399 MiB1,594 MiByes111 s9.563GB
Hetzner CX334 vCPU · 8 GB · x86_64111,316 MiB1,732 MiByes50 s9.563GB
Hetzner CX438 vCPU · 15 GB · x86_64111,101 MiB1,870 MiByes29 s9.563GB
Hetzner CAX112 vCPU · 4 GB · aarch64111,562 MiB1,564 MiByes99 s10.14GB
Hetzner CAX214 vCPU · 8 GB · aarch64111,318 MiB1,712 MiByes52 s10.14GB

Measured by AI Hosting Lab on 2026-11-05 · 5 servers · method · raw data

Hetzner prices from the API

PlanCPUvCPURAMDiskPer month excl. VATPer month incl. 21% VATPer hour incl. VAT
CAX11arm24 GB40 GB5.99 €7.25 €0.0116 €
CAX21arm48 GB80 GB10.49 €12.69 €0.0203 €
CAX31arm816 GB160 GB20.99 €25.40 €0.0407 €
CAX41arm1632 GB320 GB40.99 €49.60 €0.0795 €
CCX13x8628 GB80 GB42.99 €52.02 €0.0834 €
CCX23x86416 GB160 GB85.99 €104.05 €0.1667 €
CCX33x86832 GB240 GB138.49 €167.57 €0.2685 €
CCX43x861664 GB360 GB275.99 €333.95 €0.5352 €
CCX53x8632128 GB600 GB533.49 €645.52 €1.0346 €
CCX63x8648192 GB960 GB853.49 €1,032.72 €1.6550 €
CPX11x8622 GB40 GB5.49 €6.64 €0.0106 €
CPX02x8611 GB20 GB5.99 €7.25 €0.0116 €
CPX21x8634 GB80 GB9.49 €11.48 €0.0184 €
CPX12x8612 GB40 GB11.49 €13.90 €0.0223 €
CPX31x8648 GB160 GB17.49 €21.16 €0.0339 €
CPX22x8624 GB80 GB19.49 €23.58 €0.0378 €
CPX41x86816 GB240 GB32.49 €39.31 €0.0630 €
CPX32x8648 GB160 GB35.49 €42.94 €0.0688 €
CPX42x86816 GB320 GB69.49 €84.08 €0.1348 €
CPX51x861632 GB360 GB70.99 €85.90 €0.1377 €
CPX52x861224 GB480 GB100.49 €121.59 €0.1948 €
CPX62x861632 GB640 GB129.99 €157.29 €0.2520 €
CX23x8624 GB40 GB5.49 €6.64 €0.0106 €
CX33x8648 GB80 GB8.49 €10.27 €0.0165 €
CX43x86816 GB160 GB15.99 €19.35 €0.0310 €
CX53x861632 GB320 GB29.49 €35.68 €0.0572 €

From the Hetzner Cloud API, location hel1, checked 2026-11-05. "Incl. VAT" is what our Czech (EU consumer) account pays; businesses and buyers outside the EU usually pay the price excl. VAT. Primary IPv4 is billed separately.

Raw data

The combined file is /data/lab.json. Per-server files:

The scripts that produced them, bench.sh and n8n_load.sh, are shown in full on how we test. Data is free to reuse with credit (CC BY 4.0).

What's next in the Lab

Next up are Hostinger, DigitalOcean and Vultr on the same scripts, followed by n8n queue mode and more models. Pages that depend on a test show up only once the data is in. Start with the result most people ask about: the best VPS for n8n.

Hetzner prices on this page include 21% VAT, as charged to our Czech (EU consumer) account by the Hetzner API on 2026-11-05. Without VAT (businesses, most buyers outside the EU): CX23 5.49 €, CX33 8.49 €, CX43 15.99 €, CAX11 5.99 €, CAX21 10.49 € per month.