Contabo vs Hetzner (2026)

Updated 2026-06-14 · Independent affiliate site (disclosure)

Head-to-head comparison

Both are unmanaged European providers, so you run the operating system yourself either way. The plans below are the popular entry tiers people actually compare: Hetzner Cloud CX22 against Contabo VPS S. Specs verified on each provider's pricing page on 2026-06-14.

The pattern is consistent. Contabo wins every "how much do I get" row (RAM, cores, storage, bandwidth, datacenter count). Hetzner wins every "how fast and reliable is it" row (IOPS, TTFB, provisioning, support, uptime, automation). That single trade-off decides the whole comparison.

How we tested

We ran both plans side by side for 90 days on real workloads, not synthetic marketing benchmarks. The Hetzner CX22 ran a production WordPress site; the Contabo VPS S ran a scraping and batch pipeline plus a duplicate WordPress install for fair comparison.

• Disk: NVMe random read IOPS measured with fio (4k blocks, queue depth 32) on a fresh server, repeated across three different days to catch noisy-neighbor variance.
• Latency: TTFB measured from London, Frankfurt, and New York via WebPageTest, sustained over 7 days. Reported figures are medians.
• Provisioning: timed from clicking "create" to a server that accepts SSH. Contabo timed across two separate new orders.
• Reliability: uptime tracked over 90 days via UptimeRobot at a 1-minute interval, plus disk latency spikes logged with iostat.
• Support: we opened one real ticket on each provider and recorded time-to-first-reply.

When Hetzner wins

Hetzner is the default pick for most people, and the reasons are measurable.

• Disk speed. ~38,000 random read IOPS on the CX22 vs ~12,000 on the Contabo VPS S. WordPress, Postgres, MySQL, and anything that touches disk on every request feels this. Slow disk shows up as slow admin pages and slow queries under load.
• Latency. 8 to 12 ms TTFB from London to Falkenstein, with a tight P95. Contabo's TTFB is higher and more variable because the price point attracts denser packing and more noisy neighbors.
• Provisioning. Under 60 seconds via the console or Cloud API. Contabo can take 1 to 24 hours, particularly on first orders that hit manual review. If your workflow spins servers up and down, this gap is painful.
• Automation. A real Cloud API plus an official Terraform provider. Infrastructure-as-code on Hetzner is a solved problem; on Contabo it is not.
• Reliability and support. 99.99% measured uptime and ticket replies in hours. Contabo sat at 99.9% with occasional slower-disk windows, and support replies were slower.

When Contabo wins

Contabo is not a worse provider, it is a different deal. The whole proposition is more RAM and cores per euro, and that genuinely wins for a real set of jobs.

• Raw capacity per euro. 8 GB RAM and 4 vCPU for EUR 5.50/mo is hard to beat. If your bottleneck is "I need more cores and memory and I don't care about millisecond latency," Contabo gives you almost double the resources for one euro more.
• Throughput batch work. Compile jobs, CI pipelines, video transcoding, and large scrapes care about total cores and memory, not per-request latency. Contabo's extra capacity directly shortens these jobs.
• Storage and bandwidth headroom. 75 GB NVMe and 32 TB traffic give you room for log-heavy, data-heavy, or download-heavy work without paying for the next tier.
• Lab and self-hosted servers. A home-lab clone, a media server, or a personal tooling box benefits from cheap cores and RAM, and nobody is waiting on a page to render.

The honest limit: the moment a human is waiting on a response, Contabo's slower disk and higher latency cost you more than the euro you saved.

Our recommendation by use case

Skip the deep dive and jump to your use case.

The clean rule: build farms, CI, and scrapers go to Contabo; WordPress, apps, ecommerce, and anything user-facing goes to Hetzner. If you are unsure, default to Hetzner, because a fast server can do batch work acceptably, but a slow server cannot do user-facing work acceptably.

Honesty disclosures

We use Hetzner's referral program. If you sign up via our Hetzner link, you get EUR 20 in account credit when you spend EUR 10, and we get EUR 10 in credit after that. It is a mutual-credit referral, not a one-way commission, and the price you pay is identical. We disclose it this way because it is the most honest framing of how we monetize Hetzner.

We are not a Contabo affiliate. The Contabo link on this page points to contabo.com directly with no tracking parameters. We do not earn a commission on Contabo. When we recommend it for build farms, CI, or scrapers, that recommendation is purely on merit, we make nothing if you sign up.

Pricing reality check

The headline numbers look like Contabo wins on value: EUR 5.50/mo for 8 GB RAM and 4 vCPU against Hetzner's EUR 4.50/mo for 4 GB and 2 vCPU. Per gigabyte of RAM and per core, Contabo is cheaper, and that is real. The question is whether you can use those extra resources, or whether they sit idle behind a slower disk and noisier network.

For batch work you can use every core, so Contabo's price-per-core is the right metric and Contabo wins. For a website serving one request at a time, the relevant metric is how fast that single request completes, and there the faster Hetzner disk and lower latency mean the CX22 finishes more pages per second despite having fewer cores. We measured a duplicate WordPress install respond noticeably faster on the CX22 under the same traffic, even though the Contabo box had double the RAM. Idle RAM does not speed up a page; fast disk does.

Two cost details that surprise people. First, Hetzner now bills a small fee per IPv4 address (around EUR 0.50/mo), so an IPv6-only server is the cheapest option. Second, both providers offer larger plans with the same character: if you outgrow these tiers, the trade-off does not change, Hetzner scales up as the fast option and Contabo scales up as the high-capacity option.

Migrating between the two

Because both are unmanaged Linux providers, moving in either direction is the same job: provision the new server, copy your stack, repoint DNS. There is no proprietary panel lock-in on either side. If you are leaving Contabo because pages feel slow, plan for the DNS cutover to be the only downtime, since the Hetzner side provisions in under a minute. If you are moving to Contabo for a batch fleet, remember to order a day ahead because provisioning can take up to 24 hours on a fresh account.

A common pattern we like: run the user-facing site on Hetzner and the heavy background jobs on Contabo, connected over a private overlay or plain SSH. You get Hetzner's speed where users feel it and Contabo's cheap cores where they do not. That split is often better than forcing one provider to do both jobs.

FAQ

Which is better, Contabo or Hetzner?

For anything latency-sensitive or user-facing (WordPress, web apps, APIs, dashboards), Hetzner is better: roughly 3x faster disk (~38,000 vs ~12,000 IOPS), lower TTFB, instant provisioning, and 99.99% uptime. Contabo is the better pick only when you need maximum RAM and cores per euro for non-latency workloads like build farms, CI runners, and scrapers.

Is Contabo cheaper than Hetzner?

On raw specs per euro, yes. Contabo VPS S is EUR 5.50/mo for 8 GB RAM, 4 vCPU, 75 GB NVMe, 32 TB traffic. Hetzner CX22 is EUR 4.50/mo for 4 GB RAM, 2 shared vCPU, 40 GB NVMe, 20 TB traffic. Contabo gives double the RAM and cores for one euro more, but its per-core speed, disk, and reliability are lower, so the price only wins on bulk batch work.

Why is Hetzner faster than Contabo?

Disk and network. We measured ~38,000 random read IOPS on a fresh Hetzner CX22 vs ~12,000 on a Contabo VPS S, which matters for database-backed sites. TTFB from London to Hetzner Falkenstein is 8 to 12 ms, while Contabo runs hotter with more noisy neighbors at the price point. Hetzner also provisions instantly versus Contabo's 1 to 24 hours.

Is Contabo a good Hetzner alternative?

Only for specific workloads. If you need lots of RAM and cores at the lowest price and the work is not latency-sensitive (compile jobs, transcoding, scraping, batch processing, lab servers), Contabo is a reasonable Hetzner alternative. For production websites, ecommerce, or APIs, Hetzner is the safer alternative because disk speed, latency, and uptime are more consistent.

How long does Contabo take to provision a server?

Contabo provisioning ranges from 1 to 24 hours, especially for first-time accounts that go through manual review. Hetzner provisions a Cloud server in under 60 seconds via console or API. If you spin servers up and down often, or need infrastructure-as-code automation, Hetzner's instant provisioning is a real operational advantage.

Does Contabo or Hetzner have better support?

Hetzner support is faster and more consistent in our experience, usually resolving tickets within hours during EU business hours. Contabo support has historically been slower, with multi-day replies during busy periods and limited live chat. Neither is managed hosting, both are unmanaged providers where you run the operating system yourself.

Can I run WordPress on Contabo?

You can, but Hetzner is the better WordPress host of the two. WordPress is database-heavy, and the slower Contabo disk shows up as higher TTFB and slower admin under load. On Hetzner CX22 with Redis or LiteSpeed caching you can comfortably serve a small-to-mid WordPress site. Use Contabo for WordPress only on a staging or low-traffic internal site where latency does not matter.

Related Guides

• Best Cheap VPS Hosting (2026), the full field including both of these providers ranked by use case.
• Hetzner vs DigitalOcean, price-performance leader against the docs-and-ecosystem favorite.
• Hetzner vs Azure, the same EUR-4.50 server against enterprise cloud pricing.
• Best Windows VPS, for RDP, .NET dev, and MT5 trading with licensing-aware deployments.