Self-Hosted LLM on a $5 VPS in 2026: What Actually Works

DEV Community · HostingSift · 2026-08-12 개발(SW)

“Run your own ChatGPT for five bucks a month” sounds like cheap clickbait. And mostly it is. But the gap between clickbait and reality has narrowed a lot in 2026. Quantized 3B and 7B models have become genuinely useful. VPS providers now pack 8 GB of RAM into plans under $6. If you pick carefully, you can run a respectable local model on a budget server.

This article is about what actually works, what falls over, and which provider from our hosting comparison gives you the most compute per dollar for AI workloads.

The short version

You will not run GPT-4-class models on a $5 VPS. Stop expecting to.
You can run Llama 3.2-3B, Qwen 2.5-7B, or Mistral-7B in 4-bit quantization if you have at least 8 GB of RAM.
CPU inference is slow (5 to 15 tokens per second) but usable for personal automation, chatbots with low traffic, or embedding generation.
The providers that matter for this use case: Contabo and Netcup, both German-speaking and both useful for EU data residency. Hetzner has pulled its entire budget line from sale, so its entry point here is now €35.49 rather than €8.49.
Avoid US hyperscaler $5 plans (DO, Vultr, Linode) for LLM work. Same price, one quarter of the RAM.

Why self-host an LLM at all?

Three reasons keep coming up in our research:

Privacy. 44% of organizations cite data privacy as the top barrier to adopting hosted LLM APIs. Self-hosting means your prompts, documents, and customer data never leave your box. For law firms, clinics, and anyone building internal knowledge bases, this is non-negotiable.

Cost at volume. Hosted API costs add up fast once you run batch jobs, background summarization, or embedding pipelines. The rough break-even with a $5 VPS sits around two million tokens per day for small models. Below that, APIs are cheaper. Above that, self-hosting starts to win.

Learning and control. You get to pick the model, tune the system prompt at the weight level (via LoRA adapters), and run without rate limits. Worth it for developers.

What “self-hosted LLM” actually needs

Before we compare VPS plans, a quick hardware reality check. On CPU-only servers (which every budget VPS is), the bottleneck is RAM bandwidth, not CPU speed. Model weights must fit in RAM, and every token generation pass reads the entire weight file. So the two numbers that matter are:

RAM size. Determines which model you can load at all.
RAM bandwidth. Determines how many tokens per second you get.

Here is roughly what each tier can do in 4-bit quantization (GGUF format, llama.cpp inference):

Update (12 August 2026) — corrections. A data audit found two errors in the earlier version of this article, and we would rather say so plainly than quietly swap the numbers.

Hetzner’s entire budget line is no longer sold. The CX23/33/43/53 and CAX11/21/31/41 plans are marked “not available” on Hetzner’s own site, in every location. Their listed prices are still accurate, which is exactly why our checks missed it — we were tracking price, not stock. The cheapest Hetzner plan you can actually buy with 8 GB of RAM is now the CPX32 at €35.49, roughly four times the €8.49 CX33 this article originally recommended. That single fact changes the shape of this comparison, so the Hetzner section below has been rewritten rather than repriced.

Our Contabo prices were wrong, and the catalogue was stale. The Contabo plans quoted earlier — Cloud VPS 10 at €3.60 and Cloud VPS 20 at €5.60 — no longer exist. Contabo replaced that line with Cloud VPS 4 through 18 (the number is the vCPU count) and our scraper had been serving cached figures rather than reading the live site, understating some prices by up to 38%. Everything below is now read from Contabo’s live pricing page.

Netcup’s VPS 1000 ARM G11 did move honestly, from €6.26 to €7.77. The full change history for all three is on our Hetzner, Netcup and Contabo price history pages.

RAM Largest usable model Realistic tokens/sec (CPU) Good for 2 GB Qwen 2.5-0.5B, Llama 3.2-1B 20-40 tok/s Embeddings, simple classification 4 GB Llama 3.2-3B, Phi-3.5 Mini 8-15 tok/s Personal chatbots, summarization 8 GB Qwen 2.5-7B, Mistral-7B, Llama 3.1-8B 4-8 tok/s Production-adjacent use, RAG 16 GB Mixtral 8x7B (barely), Qwen 2.5-14B 2-4 tok/s More complex reasoning

Below 8 GB you are limited to small models. Above 8 GB you unlock the “actually useful” tier. This matters enormously when picking a VPS.

The $3 to $6 VPS plans that can run real LLMs

Pulled directly from our database, filtered to plans where the RAM is at least 8 GB (the practical floor for running a 7B model in 4-bit quantization) and that the provider will actually sell you today. The first price is the 24-month rate; the figure in brackets is what you pay month-to-month with no commitment.

Provider Plan Price / mo RAM vCPU Storage Bandwidth Contabo Cloud VPS 4 €4.40 (€5.50) 8 GB 4 vCPU 100 GB SSD Unlimited Contabo Cloud VPS 6 €6.00 (€7.50) 12 GB 6 vCPU 200 GB SSD Unlimited Netcup VPS 1000 ARM G11 €7.77 8 GB 6 vCPU (ARM) 256 GB NVMe Unlimited Contabo Cloud VPS Plus 4 €10.80 (€13.50) 8 GB 4 vCPU (AMD EPYC) 150 GB NVMe Unlimited Hetzner CPX32 €35.49 8 GB 4 vCPU 160 GB NVMe 20 TB

Compare that to the “famous” $5 cloud plans:

Provider Plan Price / mo RAM vCPU DigitalOcean Basic Regular $4.00 512 MiB 1 vCPU Vultr CC Regular $2.50 512 MB 1 vCPU Linode Shared Nanode $5.00 1 GB 1 vCPU

That 8x RAM difference at the same price point is the whole reason European providers win this category. DO, Vultr and Linode are excellent for web apps and small services. They are not designed for the RAM-heavy workload an LLM needs.

Provider deep dive

Contabo: the price-per-GB winner

Contabo Cloud VPS 4 gives you 8 GB RAM, 4 vCPU and 100 GB SSD for €4.40 a month on a two-year term, or €5.50 with no commitment. Nothing else in this comparison matches that. The catch is real: Contabo is known for inconsistent CPU steal on noisy neighbours, and network latency can be variable. For LLM inference, which is RAM-bound rather than network-bound, that tradeoff is usually acceptable.

Go up one tier to Cloud VPS 6 at €6.00 (€7.50 monthly) for 12 GB RAM and 6 vCPU. This is the sweet spot for running Qwen 2.5-14B in 4-bit while still having headroom for a web server and a small vector database on the same machine.

One note on storage, since it changed with the new line: the standard Cloud VPS tiers ship SSD, not NVMe. If disk speed matters to you — loading an 8 GB model off disk on every restart, for instance — Cloud VPS Plus 4 at €10.80 keeps the same 8 GB of RAM but moves you to 150 GB of NVMe on AMD EPYC cores.

Hetzner: no longer competing at this price

Hetzner used to be the obvious quality pick here, and this article said so. That is no longer true, for a reason that has nothing to do with price: the whole cost-optimised line — CX23 through CX53 and CAX11 through CAX41 — is marked “not available” on Hetzner’s site, in every location. The prices are still published, which is what fooled our tracker for a month, but you cannot buy the machines.

What remains purchasable with 8 GB of RAM starts at the CPX32, €35.49 a month. It is a genuinely good server — AMD EPYC cores, 160 GB NVMe, 20 TB of bandwidth, datacentres in Germany, Finland and Virginia, and a control panel that is minimal in the good way. It is simply in a different price bracket, four times the CX33 it replaces, and it is hard to justify for a hobby inference box when Contabo sells 8 GB for €4.40.

The loss worth mourning is the ARM line. The CAX21’s Ampere Altra cores benchmarked roughly 15 to 25 percent faster than the x86 CX33 on llama.cpp thanks to higher sustained memory bandwidth, which made it the best tokens-per-euro option in this whole comparison. If Hetzner brings that line back, it goes straight back into this table — our Hetzner price history page tracks it nightly.

Netcup: the quiet competitor

Less known outside the German-speaking market. The VPS 1000 ARM G11 at €7.77 gives you 8 GB RAM, 6 vCPU, and 256 GB of NVMe, which is the largest disk in this comparison. Useful if you want to store multiple models and switch between them without re-downloading 4 to 8 GB each time.

What you will actually run on 8 GB

Here is the honest menu for an 8 GB VPS with Ollama or llama.cpp:

Qwen 2.5-7B-Instruct (Q4_K_M). Best general-purpose model in this size. Strong reasoning, multi-language, clean code.
Llama 3.1-8B-Instruct (Q4_K_M). Ubiquitous, well-supported, safe default. Slightly weaker than Qwen on code.
Mistral-7B-Instruct-v0.3 (Q5_K_M). Older but predictable. Good for summarization.
Phi-3.5-Mini (Q6_K). Smaller (3.8B), fits in 4 GB of RAM, quick for chat.
nomic-embed-text or bge-small-en. Embedding models for RAG pipelines. Run these alongside the main model.

What you will not run well: anything 14B+, models with vision capabilities (qwen-vl, llava), or anything demanding low-latency streaming to many users.

Practical setup in 15 minutes

Rough sketch so you know what you are signing up for. This assumes Ubuntu 22.04 or 24.04 on a 8 GB VPS.

# 1. Install Ollama (handles download, quantization, serving)
curl -fsSL https://ollama.com/install.sh | sh

# 2. Pull a model
ollama pull qwen2.5:7b-instruct-q4_K_M

# 3. Test
ollama run qwen2.5:7b-instruct-q4_K_M "Summarize the French Revolution in two paragraphs."

# 4. Expose the API (careful with firewall rules)
# Ollama serves on localhost:11434 by default. Reverse-proxy with Caddy or Nginx.

Add Open WebUI if you want a ChatGPT-like browser interface. Add llama-swap if you want to load different models on demand without restarting.

When NOT to self-host

A $5 VPS gives you about 5 to 8 tokens per second on a 7B model. That is fine for one user typing in a chat. It is not fine for:

Real-time voice assistants. Latency will be embarrassing.
Multi-user chat services. Four concurrent users will queue and feel broken.
Anything that needs tool use, function calling at scale, or long-context (>8K tokens).
Production RAG over millions of documents with tight latency SLAs.

For those workloads, you want a GPU instance. That immediately pushes you above $100/month with Vast.ai, RunPod, or a dedicated server with a consumer GPU. At that point, OpenAI API or Anthropic API is often cheaper up to surprisingly high volumes.

Quick comparison: which $5 to $7 VPS wins?

Use case Best pick Why Lowest price with 8 GB RAM Contabo Cloud VPS 4 €4.40 for 8 GB is unmatched Best tokens per second per euro Netcup VPS 1000 ARM G11 ARM memory bandwidth, now that Hetzner’s CAX line is unavailable Largest model storage Netcup VPS 1000 ARM G11 256 GB NVMe stores a dozen models easily Fast disk for model loading Contabo Cloud VPS Plus 4 150 GB NVMe on EPYC; standard Cloud VPS is SSD Running LLM + web app on one box Contabo Cloud VPS 6 12 GB RAM leaves room for Postgres and nginx Reliability regardless of price Hetzner CPX32 The cheapest 8 GB Hetzner you can still buy, at €35.49

You can also compare them side by side on Contabo vs Hetzner or review the full cloud VPS category.

What changes this year

Two developments worth tracking in 2026:

Smaller models keep getting smarter. Phi-3.5 Mini already outperforms last year’s 13B models on some benchmarks. Expect 2B and 3B models to be genuinely production-useful by Q4 2026. That makes the 2 GB RAM tier (under $2) a legitimate option.

ARM inference is closing the gap. Apple showed what M-series can do. Ampere Altra and AWS Graviton are bringing similar memory bandwidth advantages to $5 VPS plans. If you have not tried ARM yet, this is the year.

The verdict

If you want to learn, build a personal assistant, run background summarization, or stand up a private chatbot for a small team, a €5 to €8 VPS with 8 GB of RAM is genuinely enough in 2026. Pick Contabo for raw price, Netcup for ARM throughput and storage. Hetzner is still the reliability benchmark, but until its cost-optimised line comes back, entry is €35.49 rather than €8.49. Avoid the US hyperscaler budget tiers; they are optimised for a different use case.

If you want to serve real users at real volume, stop pretending a shared VPS is the answer. Use an API, or skip straight to a GPU host.