Affiliate Program · Recommend Sunucun and earn 20% commission on every sale · 20% commissionJoin now
language-tr
menu-open
Ollama Server

Ollama server: run your own language model

Ollama lets you download and run open-source large language models (LLMs) with a single command. Install Ollama on your high-RAM NVMe cloud server and let your team reach your models from anywhere through the API or the Open WebUI interface.

  • 16 GB – 128 GB RAM
  • All-Flash NVMe, fast model loading
  • OpenAI-compatible API
  • Chat interface with Open WebUI
  • Data stays in Turkey
  • Fixed monthly price
See plans Talk to an expert From 2,345.90 ₺ / mo*billed yearly · total 28,150.83 ₺
ollama list ollama list Ubuntu 24.04 · CLD-12 ● Running llama3.1:8b 4.9 GB · loaded qwen2.5:14b 9.0 GB · loaded gemma3:4b 3.3 GB · loaded
API readylocalhost:11434
Open WebUITeam chat
  • Istanbul Tier IIIMARS data center
  • Free DDoS protectionLayer 3-4, 2.8 Tb/s
  • Weekly backupsFree, automatic
  • 24/7 supportPhone · ticket · live chat
Recommended plans

Recommended cloud servers for Ollama

The model file is loaded into RAM. We recommend at least 1.5–2 times the model size in RAM.

CLD-9
3B–8B models, single user
2,345.90₺ / mo
Billed yearly · total 28,150.83 ₺
Billed monthly 2,760.18 ₺ / mo
  • CPU4 cores · Xeon Platinum 8173M
  • RAM16 GB DDR4
  • Disk200 GB All-Flash NVMe
  • Operating systemWindows Server or Linux
Order now
CLD-10
8B models, comfortable context
3,499.75₺ / mo
Billed yearly · total 41,996.98 ₺
Billed monthly 4,117.65 ₺ / mo
  • CPU6 cores · Xeon Platinum 8173M
  • RAM24 GB DDR4
  • Disk300 GB All-Flash NVMe
  • Operating systemWindows Server or Linux
Order now
Recommended
CLD-12
Models up to 14B
5,465.56₺ / mo
Billed yearly · total 65,586.73 ₺
Billed monthly 6,430.37 ₺ / mo
  • CPU10 cores · Xeon Platinum 8173M
  • RAM32 GB DDR4
  • Disk500 GB All-Flash NVMe
  • Operating systemWindows Server or Linux
Order now
CLD-15
27B–32B models
10,465.56₺ / mo
Billed yearly · total 125,586.73 ₺
Billed monthly 12,312.72 ₺ / mo
  • CPU16 cores · Xeon Platinum 8173M
  • RAM64 GB DDR4
  • Disk1000 GB All-Flash NVMe
  • Operating systemWindows Server or Linux
Order now

*The large figure is the monthly amount when billed yearly (12 months); below it are the yearly total and the monthly billing price. Prices in Turkish lira (₺) are updated automatically from our order system; monthly, 3, 6 and 12-month billing is available. Our servers have no GPU; Ollama runs models on the CPU.

What is Ollama, and why run it on your own server?

Ollama is an open-source model runner that downloads and runs open models such as Llama, Qwen, Gemma, Mistral, Phi and DeepSeek with the ollama run model_name command. You can try models on your own computer, but to use them as a team, reach them 24/7 and connect them to your applications you need an always-on server.

Ollama on a server provides an OpenAI-compatible API, so many applications and libraries that support the OpenAI API can be pointed at your own model with a small setting change. Your prompts and documents never leave your server.

Setup summary

  1. Connect over SSH to your cloud server running Ubuntu LTS.
  2. Install Ollama with the official script: curl -fsSL https://ollama.com/install.sh | sh
  3. Download your first model: ollama pull llama3.1:8b
  4. For an interface, install Open WebUI with Docker and publish it over HTTPS behind a reverse proxy (Nginx).

Security note: the Ollama API has no authentication by default. Do not expose port 11434 directly to the internet; restrict access with a reverse proxy, a password or a VPN.

Model guide

Popular models and their RAM needs

Approximate values for 4-bit quantized versions. Long context and simultaneous requests need more RAM.

ModelSize (4-bit)Recommended RAMGood at
gemma3:4b≈3 GB8 GB+Fast answers, classification
llama3.1:8b≈5 GB16 GBGeneral assistant, summarization
qwen2.5:7b≈5 GB16 GBMultilingual text, code
qwen2.5:14b≈9 GB32 GBMore consistent long answers
deepseek-r1:14b≈9 GB32 GBStep-by-step reasoning
gemma3:27b≈17 GB64 GBHigh quality, slower answers

Model sizes follow the default tags in the Ollama library and may change between versions.

Use cases

What can you do with an Ollama server?

Internal ChatGPT

User accounts, chat history and document upload with Open WebUI.

n8n agents

Connect the n8n AI Agent node to your Ollama model.

RAG knowledge base

Vectorize and search your documents with embedding models.

Code assistant

Local code completion and review for your development team.

Translation and summaries

Summarize or translate long texts in bulk.

Model trials

Compare different models on the same server and pick the best fit.

Frequently asked questions

Common questions

Is Ollama free?

Ollama is open source and free. Each model you run has its own license; check the model license before commercial use.

Does Ollama work without a GPU?

Yes. Ollama also runs models on the CPU. Small and medium quantized models are usable on a strong processor with enough RAM; large models take longer to answer.

Which plan should I choose?

We recommend CLD-9 with 16 GB RAM for models up to 8B, CLD-12 with 32 GB RAM for 14B models, and plans with 64 GB RAM or more for 27B–32B models.

Is it safe to expose Ollama to the internet?

The API has no authentication by default. Instead of opening the port directly, use a reverse proxy with HTTPS, password protection or a VPN.

Can I run DeepSeek?

You can run distilled versions of DeepSeek-R1 such as 7B, 8B and 14B. Full-size models need hundreds of GB of memory and are not practical on a CPU server.

How do I connect my application to Ollama?

Ollama provides an OpenAI-compatible API. Point the API address in your application to your server and the model name to your Ollama model.

We are here to help

Start your own LLM server today

Write to us for help with model choice and resource planning. Our team is with you 24/7 by phone, live chat and WhatsApp.