Can a VPS Run an LLM? The Honest Math
Yes — CPU inference via Ollama runs quantized models at 5–15 tokens per second, fine for chat, automation and APIs. RAM is the deciding resource:
| Model class | Examples | RAM needed | VPS tier |
|---|---|---|---|
| 3–4B | Llama 3.2 3B, Qwen 2.5 3B | ~4 GB | Entry 4 GB |
| 7–8B | Llama 3.1 8B, Mistral 7B | 8 GB | ₹5,900/mo (8 GB) |
| 13–14B | Qwen 2.5 14B class | 16 GB | Business tier |
| 30B+ | Mixtral-class and up | 32 GB+ | Pro/dedicated |
Q4 quantization is the standard: a 7B model drops from ~14 GB to 4–5 GB of weights. A smaller model running fast beats a larger one crawling. Need GPU-class speed instead? See GPU VPS options.
Model Sizes vs RAM Required
On Ollama, the honest summary is that defaults matter more than brand names — a well-set budget plan outranks a mis-tuned premium one. WebJi® documents its defaults openly and adjusts them per ticket when workloads need it.
Step 1: Install Ollama
Install Ollama. Do this on one site or a staging copy first, then check the public URL in a private window so cache does not hide the result. Stop if the page errors, and email support@webji.services with the step name and the exact error before you repeat it on every account. WebJi® support mail is staffed 24/7 in Hindi and English; the phone line +91 141 4941811 is business hours only.
Step 2: Pull and Run Models
Pull and Run Models. Do this on one site or a staging copy first, then check the public URL in a private window so cache does not hide the result. Stop if the page errors, and email support@webji.services with the step name and the exact error before you repeat it on every account. WebJi® support mail is staffed 24/7 in Hindi and English; the phone line +91 141 4941811 is business hours only.
Step 3: Open WebUI and API Access
Two practical notes on 7B and 3B model families (Llama, Mistral, Qwen): it interacts directly with caching and database response, and misconfiguration is the most common cause of 'hosting is slow' tickets. WebJi® support checks it as part of standard migrations.
Costs: VPS per Month vs API per Million Tokens
Self-hosted on CPU: an 8 GB VPS at roughly ₹2,500–₹6,000 a month — WebJi®'s cloud entry sits at ₹5,900 with 8 GB RAM and India-located storage — serves unlimited tokens for you and a small team. Flat cost, zero per-token billing, data never leaves your server.
The API route: frontier models bill per million input tokens, with output tokens priced in multiples — cheap at low volume, brutal once a production workload or an agent loop runs daily. The crossover: sustained usage above a few million tokens a month makes the flat VPS line cheaper, and privacy-sensitive workloads make it mandatory regardless. Start on a standard VPS server plan; step up to a GPU VPS only when token speed becomes the bottleneck.
Frequently Asked Questions
Can I run an LLM on a VPS without a GPU?
For “Can I run an LLM on a VPS without a GPU”: Q4 quantization is the standard: a 7B model drops from ~14 GB to 4–5 GB of weights. A smaller model running fast beats a larger one crawling. Need GPU-class speed instead? See GPU VPS options.
How much RAM do I need to run a 7B model?
Run an LLM on a VPS (Ollama Guide) is quoted per requirement on this page, in INR, with 18% GST on the invoice. Email support@webji.services or open a ticket at https://billing.webji.in/submitticket.php?step=2&deptid=10 and the desk sends the figure before you pay. There is no dollar price converted at checkout, and a design or SEO quote is a project price — it is not a hosting renewal.
How do I install Ollama on a server?
Run an LLM on a VPS (Ollama Guide). Email support@webji.services is 24/7 in Hindi and English, average first reply under 60 minutes. Phone +91 141 4941811 is business hours. If the question is about your own account, open a ticket at https://billing.webji.in/submitticket.php?step=2&deptid=10.
Which LLMs run well on CPU only?
Run an LLM on a VPS (Ollama Guide). Email support@webji.services is 24/7 in Hindi and English, average first reply under 60 minutes. Phone +91 141 4941811 is business hours. If the question is about your own account, open a ticket at https://billing.webji.in/submitticket.php?step=2&deptid=10. The question here is: Which LLMs run well on CPU only.
How fast is CPU inference on a VPS?
Yes — CPU inference via Ollama runs quantized models at 5–15 tokens per second, fine for chat, automation and APIs. RAM is the deciding resource:
Is self-hosting an LLM cheaper than API calls?
The phone number is +91 141 4941811. Call it in business hours. It is not staffed as a 24/7 helpline, and a 2 a.m. server problem will not be picked up there. For round-the-clock help, email support@webji.services or open a ticket at https://billing.webji.in/submitticket.php?step=2&deptid=10. Email is answered in Hindi and English, usually within 60 minutes.
Is WebJi® an Indian company?
Yes. WebJi® is headquartered in Jaipur, Rajasthan and has operated since 2005. Billing, support and governance all run from India.
Do you provide GST invoices?
Every purchase and renewal gets a GST invoice with our GSTIN, and hosting prices are quoted exclusive of 18% GST. Claim input credit as applicable.