iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
There is no single best VPS for Ollama. For most people running quantized 7B or 8B models, the best value is a Linux VPS with 8 GB of RAM at minimum and 16 GB if you want reliable headroom, longer context, or 13B–14B models. Choose a plan with stronger CPUs when response speed matters; choose more RAM per dollar for asynchronous or occasional use. Models around 70B generally require 48–64 GB or more and are not practical on a standard VPS.
Roundups commonly position Contabo as a RAM-per-dollar choice and Hetzner as a CPU-performance choice, but the cited prices and specifications have not been independently verified against current provider pages. Treat them as comparison points, not quotations.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
What makes a VPS good for Ollama?
Ollama runs language models locally on the server. A VPS gives you a persistent Linux machine with remote access, but a standard VPS normally has no GPU. That means your model is loaded into system RAM and processed by virtual CPU cores.
The practical decision is therefore not simply “which provider is cheapest?” It is whether the plan has enough memory for your model, enough CPU capacity for tolerable token generation, sufficient NVMe storage for model files, and acceptable network and renewal terms. Provider marketing often omits one or more of those details.
#1 Best Overall
- Read Before You Buy — No Video Output: These adapters support charging and USB 2.0 data transfer, but cannot transmit video signals. Except for standard USB webcams (which use USB data only), they are not compatible with HDMI/DisplayPort cables, video-capable USB-C hubs, or docking stations with video output.
- Convert USB-A Ports to USB-C: Designed to connect USB-C earphones, cables, flash drives, card readers, and other USB-C accessories to standard USB-A ports. Plug-and-play with no drivers or software required.
- Aluminum Alloy Housing: Built with a sturdy aluminum alloy shell that aids in heat dissipation and protects against daily wear and scratches. Designed to maintain a stable and secure connection.
- Compact & Travel-Friendly: The ultra-compact design allows the adapter to stay plugged into your device without blocking adjacent ports or adding bulk, reducing wear and tear on your original USB ports.
- 12-Month Warranty: Backed by a 12-month manufacturer warranty for peace of mind. Designed to meet strict quality control standards for reliable everyday performance.
- RAM: determines which model sizes can load and how much context you can keep.
- CPU: largely determines token speed on a CPU-only VPS.
- Storage: must hold the model, operating system, logs, and temporary files.
- GPU access: is required for practical high-throughput inference or very large models.
- Administration: you remain responsible for Linux updates, firewall rules, authentication, and service monitoring unless the provider explicitly manages them.
How much RAM does Ollama need?
Ollama does not publish one universal RAM minimum in the official documentation results reviewed for this guide. Memory depends on parameter count, quantization, context length, concurrent requests, and the operating system. The following figures combine Ollama’s model-library guidance with independent sizing estimates, so use them as planning ranges rather than guarantees.
| Model size | Planning RAM | What the evidence says | VPS implication |
|---|---|---|---|
| 3.8B | About 4 GB | A vendor-oriented roundup estimates roughly 4 GB for Phi-3 Mini. | A 4 GB VPS may boot the model, but leaves little room for the OS or larger context. |
| 7B–8B, quantized | 8 GB minimum; 8–10 GB often cited at Q4 | Ollama’s orca-mini library page lists at least 8 GB for 7B. Another roundup estimates 8–10 GB for Q4 7B–8B models. | 8 GB is the floor. A 16 GB plan is more forgiving for context, updates, and other services. |
| 13B–14B | 16–18 GB | The orca-mini page lists at least 16 GB for 13B; another estimate puts 13B–14B Q4 models at 16–18 GB. | Prefer 24 GB if you need operating-system headroom or multiple models. |
| 70B | 48–64 GB or more | Roundups cite 48 GB or more, while Ollama’s model-library guidance lists at least 64 GB. A community guide estimates about 40 GB for a Q4 70B model, but calls the figure approximate. | Usually unsuitable for a normal VPS. Look for a dedicated high-memory or GPU machine instead. |
Quantization reduces memory use, but it does not remove the need for headroom. Context length also consumes memory. Leave capacity for the Linux kernel, Ollama itself, model metadata, and any web API or reverse proxy you run beside it. If the host starts swapping, generation can become dramatically slower or fail.
Recommended Free Tools
CPU-only Ollama: what performance to expect
CPU-only inference works well for experimentation, private assistants, low-volume APIs, and asynchronous jobs. It is less suitable when many users need immediate responses.
One roundup reports 20–40 tokens per second for 7B–8B models on CPUs, but the same source set describes that range as optimistic compared with other estimates. CPU model, sustained clock speed, thread allocation, virtualization limits, quantization, prompt length, and context size all change the result. No trustworthy cross-provider benchmark was established, so benchmark the exact plan you intend to keep.
Rank #2
- 5-in-1 USB-C Hub: Experience comprehensive connectivity featuring a Power Delivery input, two USB-A 2.0 ports, a USB-A 3.0 port, and an HDMI port. (Note: The USB-C power delivery input port is only for connecting an external wall charger to power your laptop and cannot power peripheral devices.)
- 90W Pass-Through Charging: Achieve optimal charging with 90W pass-through power to your laptop, supported by a total input of 100W, with the hub reserving 10W for operational efficiency. (Note: Wall charger not included.)
- Quick Data Transfers: Accelerate your productivity with rapid data transfers using a high-speed 5Gbps USB 3.0 port and two 480Mbps USB 2.0 ports.
- 4K HDMI Display: Enhance your visual experience with a hub capable of delivering 4K resolution at 30Hz in both mirror and extend modes. Please note that this hub is compatible with MacBook (macOS 12 and newer), Windows 10 and 11, ChromeOS, and laptops equipped with DP Alt Mode and Power Delivery. Note: This device is not compatible with Linux.
- What You Get: Anker USB-C Hub (5-in-1, 4K HDMI), welcome guide, 18-month warranty, and our friendly customer service.
When CPU-only is a sensible choice
- You need one or a few simultaneous conversations.
- Responses can take several seconds or more.
- You are running 7B–8B quantized models.
- You value low monthly cost and simple Linux deployment over maximum throughput.
When a GPU is the better answer
- You need interactive, high-throughput generation for multiple users.
- You want larger models without allocating tens of gigabytes of system RAM.
- You need predictable latency for a production application.
- You plan to use long contexts or several models concurrently.
VPS options discussed in current roundups
The figures below come from provider roundups rather than verified provider pages. Prices can change, may apply only to a region or billing term, and do not establish renewal pricing, bandwidth limits, or availability.
| Provider or plan cited | Published configuration | Reported price | How to interpret it |
|---|---|---|---|
| Contabo Cloud VPS 6 | 6 vCPU, 12 GB RAM | $7.20 per month | A 2026 roundup calls it strong RAM per dollar. The CPU is described as shared and weaker than Hetzner’s in that comparison. The figure is unverified on the provider page. |
| Contabo Cloud VPS S | 8 GB RAM | About €6 per month | Another roundup lists this approximate price. Exact CPU, storage, region, and billing conditions were not established. |
| Hetzner CPX41 | 8 vCPU, 16 GB RAM, 160 GB NVMe | €14.99 per month | Roundups associate this class with faster AMD EPYC CPU performance than the lower-cost Contabo examples. Verify current availability and renewal terms. |
| Hetzner CPX51 | 16 vCPU, 32 GB RAM, 240 GB NVMe | €19.99 per month | A higher-memory option for 13B–14B models, larger caches, or concurrent work. The cited amount is not a current provider quote. |
| Netcup plans in one pricing schema | 10–16 vCPU, 16–48 GB RAM | About €13–€32 per month | Useful as a price band for comparison, not a confirmed plan or contract price. |
| IONOS, Bluehost, Hivelocity, Cloudzy, DigitalOcean, and Vultr | Not stated consistently | Not stated | They appear in roundups, but the available material does not establish a current, apples-to-apples Ollama plan comparison. |
If your priority is maximum memory for occasional jobs, start by comparing plans in the Contabo class, then check CPU performance and storage carefully. If generation speed matters more, investigate plans in the Hetzner CPX class. Those are tendencies reported by roundups, not independent benchmark results.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteStorage and context-length planning
Community guidance suggests keeping about 50 GB of free disk space as a baseline. A model file is only part of the requirement: reserve room for the operating system, multiple quantizations, updates, logs, and temporary files. NVMe storage generally gives faster model loading than slower network-backed disks, although generation speed remains primarily CPU- or GPU-bound after the model is loaded.
Ollama’s default context length is tied to available VRAM on GPU systems:
Rank #3
- Sleek 7-in-1 USB-C Hub: Features an HDMI port, two USB-A 3.0 ports, and a USB-C data port, each providing 5Gbps transfer speeds. It also includes a USB-C PD input port for charging up to 100W and dual SD and TF card slots, all in a compact design.
- Flawless 4K@60Hz Video with HDMI: Delivers exceptional clarity and smoothness with its 4K@60Hz HDMI port, making it ideal for high-definition presentations and entertainment. (Note: Only the HDMI port supports video projection; the USB-C port is for data transfer only.)
- Double Up on Efficiency: The two USB-A 3.0 ports and a USB-C port support a fast 5Gbps data rate, significantly boosting your transfer speeds and improving productivity.
- Fast and Reliable 85W Charging: Offers high-capacity, speedy charging for laptops up to 85W, so you spend less time tethered to an outlet and more time being productive.
- What You Get: Anker USB-C Hub (7-in-1), welcome guide, 18-month warranty, and our friendly customer service.
| Available VRAM | Default context length |
|---|---|
| Under 24 GiB | 4,096 tokens |
| 24–48 GiB | 32,768 tokens |
| 48 GiB or more | 262,144 tokens |
These defaults describe VRAM tiers, not a promise that every model can use the full length at useful speed. On a CPU-only VPS, practical context limits are constrained by system RAM and latency; increasing context can consume substantial additional memory.
GPU requirements for an Ollama server
A GPU VPS is a different category from a standard virtual machine. Confirm that the provider exposes the GPU to your instance, supports the required driver stack, and permits the operating system and container configuration you plan to use.
NVIDIA on Linux
Ollama’s official Linux guidance requires NVIDIA compute capability 5.0 or newer and driver 550 or newer. GPUs with compute capability 5.0–6.2 require driver 570 or newer. Check the exact GPU model before ordering; a nominally large GPU that fails the driver or compute-capability requirement will not help.
AMD on Linux
The documented requirement is ROCm version 7. Some distributions also require the Ollama service user to belong to the render group so it can access the device.
Rank #4
- Dual Converters, Infinite Potential：Includes 2× USB C male to USB A female adapters and 2× USB A male to USB C female adapters. Perfect for a wide range of uses—tablets with Bluetooth keyboards, expand USB ports on macbook, and more. Two different converters for all your daily needs
- Next-Level 10Gbps & 3A Charging: No more slow 480Mbps, this usb to usb c adapter has a transfer speed of up to 10Gbps, allowing you to do more transferring in less time. This usb adapter fits both USB A and USB C charger, supporting up to 3A fast charging
- Upgraded Exquisite Craftsmanship: With an aluminum alloy housing and metal connector, the usbc to usb adapter is extremely durable and sturdy. Rigorously tested to withstand more than 10,000 times of plugging and unplugging, ensuring long-lasting performance
- Broad Compatible: The usb c to usb adapter widely supports all USB C/ USB A devices like laptops, tablets, cellphones, car chargers, and phone chargers. Such as compatible with MacBook Pro/Air 2023/2022, Thunderbolt 4/3 Devices,Apple MagSafe Watch 9/8/7/SE/Ultra, iPad Pro 2022/2021, Samsung Galaxy S23/S20/S10, and iPhone 17/16/15 Pro. Plug and play
- Please Note: To reach 10Gbps speed, keep the cable under 3.3 ft. For USB A Male to USB C adapters, try flipping the USB C connector. USB C Male to USB A adapters support bidirectional 10Gbps transfer within 3.3 ft
Other GPUs
Vulkan support is available for other hardware, but it may require additional components and permission or capability settings so Ollama can see the GPU’s VRAM. Treat this as a hands-on configuration rather than an automatic feature of a generic VPS.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →No current, independently verified GPU-VPS price comparison was established. One roundup gives a broad self-hosted GPU estimate of $500–$1,500 per month, but it is an unsourced indication rather than a quote or market average.
Deploy Ollama on a Linux VPS
The exact installer differs by distribution. Use a current 64-bit Linux image supported by your provider, install security updates, and follow Ollama’s official Linux installation instructions. After installation, the following checks and commands help validate a CPU deployment.
- Size the server first. Choose RAM for the model, then add room for the OS and context. For a first 7B–8B experiment, 8 GB is the minimum practical target; 16 GB is the safer general-purpose choice.
- Update the system and create an administrative path. Log in with SSH keys rather than a password, apply distribution updates, and create a non-root administrator with
sudoaccess. - Install Ollama. Follow the current Linux instructions from Ollama, then confirm the service is enabled and running with your distribution’s service manager.
- Check resources before downloading a model. Run
free -hfor memory anddf -hfor disk. Do not proceed if available disk space is only a few gigabytes above the model’s expected size. - Download a model that fits. Use
ollama pull <model-name>, replacing the placeholder with the exact library tag you selected. A quantized 7B–8B model is the sensible first test on an 8–16 GB CPU VPS. - Run an interactive test. Use
ollama run <model-name>. Record startup time, tokens per second, memory consumption, and whether the kernel reports swapping. - Measure under realistic load. Repeat with your normal prompt length and context. A short benchmark can look good while long prompts exhaust memory.
- Expose an API only after hardening the host. Keep Ollama bound to a private interface where possible, place a TLS-terminating reverse proxy in front of it when remote clients need access, and restrict inbound firewall rules to known networks.
How to choose a plan by workload
Private 7B–8B assistant
Start with 8 GB RAM and enough NVMe space for at least one model plus operating-system overhead. Choose 16 GB instead if you expect long prompts, background services, or frequent model changes.
Best Value
- 5-in-1 Connectivity: Equipped with a 4K HDMI port, a 5 Gbps USB-C data port, two 5 Gbps USB-A ports, and a USB C 100W PD-IN port. Note: The USB C 100W PD-IN port supports only charging and does not support data transfer devices such as headphones or speakers.
- Powerful Pass-Through Charging: Supports up to 85W pass-through charging so you can power up your laptop while you use the hub. Note: Pass-through charging requires a charger (not included). Note: To achieve full power for iPad, we recommend using a 45W wall charger.
- Transfer Files in Seconds: Move files to and from your laptop at speeds of up to 5 Gbps via the USB-C and USB-A data ports. Note: The USB C 5Gbps Data port does not support video output.
- HD Display: Connect to the HDMI port to stream or mirror content to an external monitor in resolutions of up to 4K@30Hz. Note: The USB-C ports do not support video output.
- What You Get: Anker 332 USB-C Hub (5-in-1), welcome guide, our worry-free 18-month warranty, and friendly customer service.
13B–14B experimentation
Use at least 16–18 GB according to the sizing estimates, with additional headroom preferred. More vCPUs may improve responsiveness, but extra cores cannot compensate for insufficient RAM.
Small internal API
Favor stronger, consistently available CPUs and 16–32 GB RAM. Test concurrent requests, because a plan that is acceptable for one conversation may queue or swap when several requests arrive together.
70B or larger
Do not force a 70B model onto a conventional low-cost VPS. The cited guidance ranges from 48 GB to 64 GB or more, and the lower community estimate is explicitly approximate. Compare dedicated high-memory or GPU systems instead.
Common failure modes and fixes
The model will not load
- Check
free -hand system logs for out-of-memory events. - Stop unrelated services and choose a smaller or more aggressively quantized model.
- Move to the next RAM tier; adding CPU cores does not solve a memory shortfall.
Generation is extremely slow
- Check whether the VPS is swapping with
free -hand your distribution’s virtual-memory tools. - Reduce context length and test a smaller model.
- Compare a plan with faster dedicated or higher-performing shared CPUs; published token-rate claims are not reliable substitutes for your own test.
Disk fills during model downloads
- Run
df -hbefore each download. - Remove unused model variants and old logs.
- Choose a larger NVMe volume rather than relying on a nearly full boot disk.
GPU is not detected
- Confirm that the provider actually passes through the GPU to the instance.
- Verify NVIDIA compute capability and driver versions, or the required ROCm version for AMD.
- Check device permissions, including the
rendergroup requirement on some AMD distributions.
The API is exposed to the internet unintentionally
- Review listening addresses and firewall rules.
- Require authentication at the reverse proxy or application layer.
- Use SSH tunnels or a private network for administrative access.
A buying checklist for 2026
- Identify the largest model you will actually run, including its quantization.
- Reserve RAM beyond the model estimate for context, Linux, and other services.
- Keep roughly 50 GB of free storage as a starting point, then add space for every additional model.
- Compare CPU architecture and sustained performance, not only vCPU count.
- Confirm whether CPUs are shared, dedicated, or subject to stated limits.
- Check region, bandwidth, IPv4/IPv6 availability, snapshots, backups, and support before ordering.
- Verify recurring and renewal pricing directly with the provider; roundup prices are not guarantees.
- For GPU plans, verify driver, compute-capability, ROCm, Vulkan, and device-permission requirements.
- Run your own workload test before committing to a long billing term.
Verdict
For a low-cost Ollama server, a Linux CPU VPS with 8 GB RAM is the entry point for quantized 7B–8B models, while 16 GB is the more comfortable default. The roundup evidence favors Contabo-style plans when RAM per dollar is the deciding factor and Hetzner CPX-style plans when CPU performance is more important, but neither conclusion is a verified benchmark or a current price guarantee. Once your target moves to 13B–14B, plan for at least 16–18 GB and additional headroom. For 70B models, long contexts, or multi-user production traffic, skip the standard VPS and evaluate high-memory or GPU infrastructure instead.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
