Contabo VPS review: value, regions, and the AI angle

Company snapshot

Contabo is a German-rooted hosting company selling self-managed virtual and dedicated servers. Its historical marketing cited more than two decades in business and hundreds of thousands of customers; those figures describe scale, not a support commitment. The buyer owns patching, application security, recovery, and capacity planning.

For residency, distinguish company address from server location. Contabo's documented EU Hub is Lauterbourg, France, near the German border. If a contract requires a specific country, verify the selected site and DPA before ordering. An EU selection alone does not settle every data-handling requirement.

This cluster concerns ordinary Cloud/Core VPS and the separate Performance family. Contabo also documents a separate GPU VPS. Core CPU capacity does not include GPU memory: a larger RAM allocation cannot satisfy a software requirement for CUDA or dedicated VRAM.

What Contabo is strong at

Contabo’s durable strength is RAM capacity at the recorded price on long-term Core VPS rates, paired with unlimited inbound and outbound traffic subject to fair use, and a global region list historically listing EU, UK, three US metros, Singapore, Japan, India, and Australia. If your workload is memory-heavy and network-chatty—self-hosted databases, n8n with modest concurrency, Nextcloud, a local LLM that lives mostly in CPU RAM—Contabo’s price board is designed to look generous next to hyperscaler “always on” instances.

Other strengths that survive honest inspection:

• Predictable monthly framing. Contabo advertises effective monthly rates for multi-month terms (often showing 12- or 24-month effective prices including VAT toggles). Record the initial-term total, effective monthly rate, renewal charge, taxes, and add-ons from Contabo checkout. Re-check these before purchase— Contabo may advertise term discounts that differ from later renewals.
• Snapshots + Auto Backup add-on. Cloud VPS plans include a plan-limited number of snapshots (30-day retention per Contabo help). Auto Backup is a paid add-on with daily off-server copies retained about 10 days. Together they form a usable safety net if you also keep an off-provider export.
• 1-click applications. Contabo lists n8n among 1-click apps (Docker + Caddy + Let’s Encrypt on Ubuntu 24.04 per Contabo help/blog). That matters for automation buyers who want a first instance without writing Compose files on day one—and who still accept that app upkeep is theirs.
• Storage type on Core vs Performance. Contabo Core is CPU-based; Contabo’s sales page and docs can disagree on whether a given Core tier offers SSD, NVMe, or both—verify the selected SSD/NVMe option and its capacity in checkout. Performance is a separate product family. For AI weight files and Docker layers, faster disks (when Contabo sells them on the SKU you buy) are the quieter life.

Compare Core RAM capacity against the complete checkout cost, including contract length, location fees, and backup add-ons. Check port limits, fair-use terms, and support expectations alongside that price—covered below.

Pricing as listed — verify on Contabo before you buy

Prices below are transcribed from Contabo’s public VPS pages as researched for this draft on 4 November 2026 (as listed on contabo.com — re-check before purchase). Contabo shows both a higher sticker and a lower “effective” monthly figure tied to long-term terms (commonly the 24-month effective rate including VAT on EU pages). Your checkout may differ by currency (EUR/USD/GBP), VAT treatment, contract length, setup fees, and location surcharges outside the EU hub.

Traffic: every plan advertises unlimited traffic with a fair-usage policy. Contabo states it may throttle exceptionally high or disruptive usage. Port speed—not a soft monthly TB cap—is the possible throttle under Contabo’s fair-use terms most buyers feel first.

Setup fees: Contabo notes a one-time setup fee can apply to new Cloud VPS 4 on monthly contracts; longer terms often waive it. Confirm in the configurator.

Location fees: advertised Core prices are for the EU region. Check the location surcharge for your selected region and plan. Stock can also show “out of stock” for a region on a given day.

GPU VPS (separate product; historical price example): The earlier research recorded a dedicated NVIDIA RTX PRO 6000 class GPU VPS (18 vCPU / 96 GB RAM / ~900 GB NVMe) with list pricing on the order of €999 / $1199 / £969 per month excluding VAT, with multi-month discounts advertised on Contabo’s blog. Available regions at launch notes: Europe and US Central. This is not the Core VPS ladder; do not assume every Contabo VPS includes a GPU.

Recompute your own cart with the AiXVPS cost calculator—paste Contabo’s live numbers; leave blanks blank.

Official site and checkout experience

Use the public catalog to shortlist, then record the actual configurator summary. Plan cards, VAT toggles, term discounts, regional stock, and image choices can change the payable total. Keep the order confirmation alongside the initial invoice so a later renewal has a clear comparison point.

Provisioning and application readiness are different milestones. Contabo's n8n help allows an installation window after ordering; an IP address arriving by email does not prove that TLS, login, or the database is ready. The setup guide covers those checks and recovery access.

Data centers and regions

As listed on Contabo’s locations materials (re-check contabo.com/locations before purchase):

• European Union — Hub Europe, Lauterbourg (FR/DE border). Base pricing; no EU location surcharge on the usual Core board.
• United Kingdom — Portsmouth. Location fee applies.
• United States East — Carlstadt / New Jersey–New York area (Contabo also says “New York” in marketing).
• United States Central — St. Louis.
• United States West — Seattle.
• Singapore.
• Japan — Tokyo.
• India — Mumbai / Navi Mumbai area.
• Australia — Sydney.

Older Contabo copy sometimes still mentions “five data centers in Germany.” Operationally, new EU Cloud VPS deployments are described as landing in Hub Europe (Lauterbourg). If residency language matters, screenshot the region you select at checkout and keep Contabo’s confirmation email.

Latency is geography: an n8n webhook farm for US users belongs in a US region even if EU is cheaper; a GDPR-first personal AI box often belongs in EU even if Singapore is empty. Price the location fee as a first-class line item, not a footnote.

Pain points versus earned praise

Earned strengths: RAM capacity, regional options, and useful recovery tools—when operated deliberately.

Pain points Contabo cannot dodge:

• Support wait. Assume ticket-based support and keep your own runbooks; do not buy Contabo expecting dedicated account-manager response times.
• Port caps vs hyperscalers. 200 Mbit/s on the entry plan is not “slow internet”—it is a shared-port product. Bulk model pulls, multi-stream video egress, or naive mirror syncs will feel it. Higher plans open 600 Mbit/s–1 Gbit/s; still not a dedicated 10G story.
• Fair-use throttle risk. Unlimited is not unregulated. Contabo reserves the right to throttle exceptional or disruptive patterns. Design pipelines that can pause, and avoid treating Contabo as a free CDN.
• Noisy-neighbor / shared Core reality. Contabo positions Core for lighter, cost-efficient workloads and Performance for sustained compute. If your local model saturates every core all day, budget for Performance or GPU VPS—or accept slower interactive tokens on Core.
• Cancel deletes snapshots/backups. Contabo’s cancel flow warns that cancelling a VPS permanently deletes associated snapshots and backups. Export first. Details live in the billing & backup article.

Honesty cuts both ways: Value depends on the full invoice and your ability to operate the workload. Include recovery time and support expectations in that decision.

Who Contabo fits

If you are still mapping the workload itself, start from AiXVPS hubs: n8n, Local AI, WordPress—then return here to place Contabo on the shortlist.

How Contabo connects to AI and automation

Contabo does not sell “managed ChatGPT on Contabo.” The AI story is buyer-built:

1. Core / Performance VPS as the durable worker — run Ollama-style stacks, embeddings services, or API gateways in CPU RAM; store weights on NVMe; keep secrets and reverse proxies under your control.
2. n8n 1-click (when listed) — orchestrate calls to your own models or external AI APIs for document pipelines, image jobs, and video post-steps. Contabo’s help describes Ubuntu 24.04, Docker, Caddy, and Let’s Encrypt for that add-on.
3. GPU VPS when VRAM is the bottleneck — Contabo’s dedicated GPU VPS SKU exists for serious single-node inference/fine-tune patterns; verify price, region (EU / US Central at documented launch), and CUDA image before you redesign your architecture around it.
4. Split-box discipline — many teams keep n8n on a mid Core plan and heavy models on a larger RAM or GPU box. Verify private networking or use authenticated TLS between boxes.

I do this: which Contabo VPS tier?

Editorial starting estimates, not Contabo guarantees. “mid Core” is shorthand here for a roughly 12–24 GB RAM planning envelope, not a formal SKU. Compare the live Cloud VPS / Core, Performance, and separate GPU VPS catalogs. Verify plan names, capacities, SSD/NVMe selection, stock, and complete price in the cart; these sizing estimates are not quotes.

RAM floor = resident weights + context/concurrency memory + runtime, OS, and application margin. For example, 6 GB of resident weights plus 4 GB for context and working buffers plus 6 GB for applications and reserve gives a 16 GB planning floor; choose a cart capacity above it and test peaks. Weight download size is not an exact resident-memory measurement. KV-cache grows with context and simultaneous requests.

Disk floor = stored weights + reserve + peak intermediates. A 20 GB model collection, 40 GB for OS, containers, logs and recovery copies, and 30 GB of media scratch space already requires 90 GB before filesystem headroom. Include temporary downloads and old/new model overlap. A larger disk cannot replace RAM; swap cannot turn CPU RAM into VRAM.

Larger resident weights may justify 48–96 GB RAM while still running on CPU. Sustained CPU demand points toward comparing Performance specifications, which still use shared vCPUs. A true VRAM requirement points toward the separate GPU VPS and its documented 96 GB GPU memory, subject to current availability. Follow local AI sizing and the CPU recipe, the automation starter, and the n8n hub for workload-specific checks.

For every size or family change, default to order new → migrate and verify → cancel old. Budget overlap, test credentials and restore paths, then switch traffic. Retain independent exports before cancellation deletes the old service's snapshots and backups.

Deep dives: Local AI on Contabo and AI automation pipelines. Editorial boundaries about labs and speed tables live on About—this review stays on fit, cost, and terms.

Procurement checklist for Contabo

1. Write the workload: n8n only, local model only, WordPress, or mixed—and whether same-box is intentional.
2. Pick region for latency and residency; add location fee into the calculator.
3. Choose Core vs Performance vs GPU VPS from Contabo’s current catalog—not from memory of a blog post.
4. Decide Core storage type/capacity in checkout (SSD and/or NVMe as Contabo lists) (separate Contabo product lines—verify), storage extension, and model disk floor (weights + reserve + intermediates).
5. Enable Auto Backup if restores matter; still keep an off-provider export of workflows and weights.
6. If ordering n8n 1-click, save the server password selected during ordering, which Contabo’s n8n image initially also uses; wait for install completion.
7. Read cancel, revocation (14-day private), and auto-renew refund (~72h, not Germany) articles on help.contabo.com before the first renewal.
8. Paste live cart figures into the cost tool.

Checkout line items to record

Before Contabo charges the first invoice, copy these fields into your notes: plan name (Cloud VPS 4–18 or Performance/GPU), region, storage type and capacity as shown in the cart, port speed, included snapshots, Auto Backup yes/no and fee, OS/image, 1-click apps, contract length, currency, VAT treatment, initial-term total, and the renewal amount Contabo displays if separate from the intro term.

Save the cart and renewal expectation with the invoice; a public price board does not preserve your purchase terms.

For AI buyers specifically, add: assumed weight size M, disk reserve R, whether n8n shares the box, and the off-site export destination. Contabo will not remind you that model files and workflow JSON are your problem when cancel deletes Contabo-side backups.

Where to go next

Contabo setup guide Local AI on Contabo AI automation Billing & backup Cost calculator

Sources and updates

• Contabo Cloud / Core VPS product pages — contabo.com/en/vps/ (prices, ports, traffic fair use) — re-check before purchase.
• Contabo locations — locations.
• n8n on Contabo help/blog — 1-click install notes.
• Contabo docs — Auto Backup; snapshot help (verify live).
• GPU VPS docs/blog — dedicated NVIDIA SKU, regions, list pricing.
• Cancel / refund / revocation articles on help.contabo.com.

Page status: Draft cluster for local preview. Last editorial pass: 2026-11-05 (America/Los_Angeles, PST). Contabo figures change—always verify on the provider site.