# Choosing a VPS for self-hosted AI: run Hermes Agent, Claude Code, n8n, Open WebUI and chatbots, on a yearly plan in a region Claude supports (2026-11)

Last price check: 2026-11-17

Agents like Hermes Agent, Claude Code, Codex and n8n, self-hosted AI suites like Open WebUI and Dify, and Telegram, Discord and Feishu bots connected to LLMs are all always-on services: they need to stay online, call the Anthropic or OpenAI API directly, and should not tie up your own computer. That makes them a natural fit for a cheap VPS. Below, by use case: how much server you need, where to put it, which billing to choose, a table of plans in stock now, and finally how to deploy.

## How much server you need

Anthropic's official minimum for Claude Code is 4GB RAM, an x64 or ARM64 processor, and Debian 10 or Ubuntu 20.04 or later; on large repositories a single session can use more than 2.5GB. An agent does not just call an API, it works on the machine: Claude Code clones repositories, installs dependencies, runs builds and tests, and opens a headless browser to look at pages, all of which use memory; with 4G, a browser open plus one build is enough to run out. We recommend 2 vCPU 8G: the official 4G is the floor, and 8G lets the agent, the browser and builds run at the same time without crowding each other; for a light agent that only calls APIs, such as Hermes Agent, 4G will do; for n8n and multi-service Docker setups, also plan on 8G. Disk 40G or more: Docker images, node_modules and browsers each take a few GB. Traffic is small, only API requests and text go back and forth, so there is no need to pay extra for traffic.

Browser agents need more memory. Agents like OpenClaw that open Chrome and click through web pages themselves, like Claude Code running Playwright tests, use a few hundred MB per tab; a session with several pages open takes one or two GB, and with the agent itself and the system on top, that is the main reason for 8G rather than 4G. To run two such agents at the same time, buy 16G.

## Put the server in a region Claude supports

Requests to the model API go out from whichever machine the agent is installed on. Anthropic only provides service in the countries and regions it supports, and neither mainland China nor Hong Kong is among them, so the server must be in a supported region such as the US, Japan or Singapore. The “Claude” column in this site's tables follows the official list of supported regions; whether you can use it is determined by Anthropic's official list and terms of use.

If you are in China and often open the agent's web panel, the server's route affects how responsive the panel feels, but not the agent's own work. For choosing a route, see [How to choose a China route for Telecom, Unicom and Mobile](https://vps.appsail.dev/guide/isp-lines/).

## 8G plans in stock: available for Claude, sorted by effective monthly price

An agent runs around the clock, so just pay yearly: the effective monthly price is often half the monthly rate, the server stays put, and you do not have to deal with renewals every month. In the table below, “Monthly price” is the yearly price divided by 12 where a yearly price exists, converted to US dollars; the “List price and billing” column is the vendor's list price and billing period, and plans whose first-year price differs from the renewal price are marked “Renewal”. Yearly plans are often limited stock; if one is out of stock, wait for a restock. This site checks stock daily. The cheapest right now is $8.79/mo. 8G is this site's Main server tier; filter “Use” by “Main server” on the home page to see them all. Click a column header to sort; for the full list and filters, see the [home page](https://vps.appsail.dev/en/).

| Vendor | Plan | Specs | Data centre | Monthly price | List price and billing | Traffic |
| --- | --- | --- | --- | --- | --- | --- |
| [Hostinger](https://vps.appsail.dev/en/hostinger/) | [KVM 2](https://vps.appsail.dev/go/hostinger/KVM%202) | 2 vCPU / 8G / 100GB NVMe | Same price in every data centre | $8.79 | $8.79/mo, 24 months prepaid (Black Friday sale, 64% off the list price of $24.49/mo)Renewal $14.99/mo | 8TB |
| [RackNerd](https://vps.appsail.dev/en/racknerd/) | [8 GB KVM VPS yearly deal](https://vps.appsail.dev/go/racknerd/8%20GB%20KVM%20VPS%20%E5%B9%B4%E4%BB%98%E7%89%B9%E4%BB%B7) | 7 vCPU / 8G / 150GB SSD | Same price in every data centre | $10.00 | $119.99/yrSame price on renewal | 20TB |
| [HostKvm](https://vps.appsail.dev/en/hostkvm/) | [Japan Japan-Gold (overseas IP, Tokyo)](https://vps.appsail.dev/go/hostkvm/%E6%97%A5%E6%9C%AC%20Japan-Gold%EF%BC%88%E6%B5%B7%E5%A4%96%20IP%EF%BC%8C%E4%B8%9C%E4%BA%AC%EF%BC%89) | 4 vCPU / 8G / 80GB | Tokyo | $15.00 | $15/mo, yearly $180; Japan IP (Osaka SoftBank) adds $5/mo | 2000GB |
| [HostKvm](https://vps.appsail.dev/en/hostkvm/) | [South Korea KR-Gold (overseas IP, includes mainland optimisation)](https://vps.appsail.dev/go/hostkvm/%E9%9F%A9%E5%9B%BD%20KR-Gold%EF%BC%88%E6%B5%B7%E5%A4%96%20IP%EF%BC%8C%E5%90%AB%E5%86%85%E5%9C%B0%E4%BC%98%E5%8C%96%EF%BC%89) | 4 vCPU / 8G / 80GB | South Korea | $15.00 | $15/mo, yearly $180; Korean local IP adds $5/mo | 1500GB |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [US CN2 Elastic 2 vCPU 8G](https://vps.appsail.dev/go/lcayun/%E7%BE%8E%E5%9B%BD%20CN2%20%E5%BC%B9%E6%80%A7%202%20%E6%A0%B8%208G) | 2 vCPU / 8G / 30GB | US West (city not stated by vendor) | $15.28 | ¥1225/yr, monthly ¥105 | not stated |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [Japan BGP Elastic 2 vCPU 8G](https://vps.appsail.dev/go/lcayun/%E6%97%A5%E6%9C%AC%20BGP%20%E5%BC%B9%E6%80%A7%202%20%E6%A0%B8%208G) | 2 vCPU / 8G / 30GB | Japan | $16.34 | ¥1310/yr, monthly ¥115 | 500GB |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [South Korea CN2 Elastic 2 vCPU 8G](https://vps.appsail.dev/go/lcayun/%E9%9F%A9%E5%9B%BD%20CN2%20%E5%BC%B9%E6%80%A7%202%20%E6%A0%B8%208G) | 2 vCPU / 8G / 20GB | South Korea | $17.66 | ¥118/mo | not stated |
| [Jtti](https://vps.appsail.dev/en/jtti/) | [USA CLOUD 4C8G 10M 6-month deal (limit 1 per customer)](https://vps.appsail.dev/go/jtti/USA%20CLOUD%204C8G%2010M%20%E5%8D%8A%E5%B9%B4%E4%BB%98%E7%89%B9%E6%83%A0%EF%BC%88%E6%AF%8F%E4%BA%BA%E9%99%90%201%20%E5%8F%B0%EF%BC%89) | 4 vCPU / 8G / 50GB | Los Angeles | $23.17 | $139/6 mo (website sale, 71% off, list price $469.98, limit 1 per customer)Same price on renewal | unlimited traffic |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [Singapore Elastic 2 vCPU 8G](https://vps.appsail.dev/go/lcayun/%E6%96%B0%E5%8A%A0%E5%9D%A1%20%E5%BC%B9%E6%80%A7%202%20%E6%A0%B8%208G) | 2 vCPU / 8G / 30GB | Singapore | $25.45 | ¥170/mo | 1000GB |
| [TmhHost](https://vps.appsail.dev/en/tmhhost/) | [US 9929 #6](https://vps.appsail.dev/go/tmhhost/%E7%BE%8E%E5%9B%BD%209929%20%236) | 4 vCPU / 8G / 60GB SSD | Los Angeles | $25.90 | ¥173/mo | 4000GB |
| [TmhHost](https://vps.appsail.dev/en/tmhhost/) | [US 4837 #6](https://vps.appsail.dev/go/tmhhost/%E7%BE%8E%E5%9B%BD%204837%20%236) | 4 vCPU / 8G / 60GB SSD | Los Angeles | $25.90 | ¥173/mo | 5000GB |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [Taiwan Elastic 2 vCPU 8G](https://vps.appsail.dev/go/lcayun/%E4%B8%AD%E5%9B%BD%E5%8F%B0%E6%B9%BE%20%E5%BC%B9%E6%80%A7%202%20%E6%A0%B8%208G) | 2 vCPU / 8G / 30GB | Taiwan | $26.65 | ¥178/mo | 1000GB |

## How big a server Open WebUI, Dify, n8n and AnythingLLM need

These are self-hosted AI suites: Open WebUI and AnythingLLM are your own chat interface, Dify builds workflows and knowledge bases, and n8n is workflow automation with LLM nodes. They do not run models themselves, they only pass requests on to the Anthropic or OpenAI API, so they need memory rather than compute: Open WebUI or n8n alone runs fine on 2 vCPU 4G; Dify brings up a dozen or so containers (PostgreSQL, Redis, a vector store, a sandbox), so plan on 8G; more if the knowledge base computes embeddings locally. The data centre must also be in a region Claude supports, because the API requests go out from this machine. The 8G table above is meant for these too; no need to buy anything else.

## What server to use for chatbots and MCP services

A Telegram, Discord or Feishu bot connected to an LLM, or an always-on MCP service or webhook receiver, only receives messages, calls an API and replies; idle memory use is a few dozen MB, so a smallest plan of 1 to 2G is enough, the cheapest tier. The requirements are the same as above: the server in a region Claude supports, running 24 hours a day, paid yearly. The table below shows the cheapest plans that are in stock now, available for Claude, with 1 to 2G, sorted by monthly price; click a column header to sort. Most of these plans come with China-optimised routes; bots do not need the route, but it does no harm.

| Vendor | Plan | Specs | Data centre | Monthly price | List price and billing | Traffic |
| --- | --- | --- | --- | --- | --- | --- |
| [LisaHost](https://vps.appsail.dev/en/lisahost/) | [9929 premium network yearly deal (non-native IP)](https://vps.appsail.dev/go/lisahost/9929%20%E7%B2%BE%E5%93%81%E7%BD%91%20%E5%B9%B4%E4%BB%98%E7%89%B9%E4%BB%B7%EF%BC%88%E9%9D%9E%E5%8E%9F%E7%94%9F%20IP%EF%BC%89) | 1 vCPU / 1G / 10GB SSD | Los Angeles | $2.48 | ¥199/yr | 200GB |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [Japan BGP Elastic, smallest plan](https://vps.appsail.dev/go/lcayun/%E6%97%A5%E6%9C%AC%20BGP%20%E5%BC%B9%E6%80%A7%E6%9C%80%E4%BD%8E%E9%85%8D) | 1 vCPU / 1G / 30GB | Japan | $4.37 | From ¥350/yr, monthly from ¥35 | not stated |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [Japan direct to all 3 carriers, deal 2 vCPU 2G](https://vps.appsail.dev/go/lcayun/%E6%97%A5%E6%9C%AC%20%E4%B8%89%E7%BD%91%E7%9B%B4%E8%BF%9E%20%E7%89%B9%E6%83%A0%E4%BA%91%202%20%E6%A0%B8%202G) | 2 vCPU / 2G / 50GB | Japan | $4.49 | ¥29.99/mo (sale price, regular price ¥69)Same price on renewal | 500GB |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [South Korea direct to all 3 carriers, deal 2 vCPU 2G](https://vps.appsail.dev/go/lcayun/%E9%9F%A9%E5%9B%BD%20%E4%B8%89%E7%BD%91%E7%9B%B4%E8%BF%9E%20%E7%89%B9%E6%83%A0%E4%BA%91%202%20%E6%A0%B8%202G) | 2 vCPU / 2G / 50GB | South Korea | $4.49 | ¥29.99/mo (sale price, regular price ¥58)Same price on renewal | 500GB |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [US CN2 Elastic 2 vCPU 1G](https://vps.appsail.dev/go/lcayun/%E7%BE%8E%E5%9B%BD%20CN2%20%E5%BC%B9%E6%80%A7%202%20%E6%A0%B8%201G) | 2 vCPU / 1G / 30GB | US West (city not stated by vendor) | $4.80 | From ¥385/yr, monthly from ¥35 | not stated |
| [LisaHost](https://vps.appsail.dev/en/lisahost/) | [4837 3-carrier return route Dual-ISP residential yearly deal](https://vps.appsail.dev/go/lisahost/4837%20%E4%B8%89%E7%BD%91%E5%9B%9E%E7%A8%8B%20%E5%8F%8C%20ISP%20%E5%AE%B6%E5%AE%BD%20%E5%B9%B4%E4%BB%98%E7%89%B9%E4%BB%B7) | 1 vCPU / 1G / 10GB NVMe | Los Angeles | $4.98 | ¥399/yr | 600GB |
| [LisaHost](https://vps.appsail.dev/en/lisahost/) | [Chicago/New York Dual-ISP residential yearly deal](https://vps.appsail.dev/go/lisahost/%E8%8A%9D%E5%8A%A0%E5%93%A5%2F%E7%BA%BD%E7%BA%A6%20%E5%8F%8C%20ISP%20%E5%AE%B6%E5%AE%BD%20%E5%B9%B4%E4%BB%98%E7%89%B9%E4%BB%B7) | 1 vCPU / 1G / 10GB NVMe | Chicago, New York | $4.98 | ¥399/yr | 600GB |
| [Lcayun](https://vps.appsail.dev/en/lcayun/) | [South Korea CN2 Elastic, smallest plan](https://vps.appsail.dev/go/lcayun/%E9%9F%A9%E5%9B%BD%20CN2%20%E5%BC%B9%E6%80%A7%E6%9C%80%E4%BD%8E%E9%85%8D) | 1 vCPU / 1G / 20GB | South Korea | $5.69 | From ¥38/mo | not stated |

## Can you run open-source models on a VPS

Not recommended. Running a 7B model in Ollama takes 8 to 16G of RAM, and a VPS without a GPU produces only a few tokens per second, a far worse experience than calling an API, and no cheaper; to run models you need to rent a GPU cloud, which is not a VPS and is not listed on this site. The right uses for a VPS are the ones above: running agents, suites and bots, and leaving the model to the API.

## How to deploy: install Debian and let Claude Code or Codex on your own computer do the rest

1. Choose Debian as the OS. It has the smallest footprint, so the same RAM leaves more room for the agent and the browser, which amounts to saving money; it is usually among the vendor's images.
2. You do not need to type commands yourself. Open Claude Code or Codex on your own computer, give it the VPS address and key, and let it SSH in to install Docker, pull the agent and set it to start on boot; when it is done, have it write down for you how to access it.
3. Put the API key in an environment variable or a `.env` file; never commit it to the code repository.
4. Do not expose the agent's web panel directly to the internet: have it listen on localhost only and reach it through an SSH tunnel, `ssh -L 5678:127.0.0.1:5678 user@server-ip`; or put a password-protected reverse proxy in front of it.
5. Open only the SSH port in the firewall, log in with a key and disable password login. Use systemd or Docker's `--restart unless-stopped` so services restart automatically after a crash.

Guides: [VPS for self-hosted AI agents](https://vps.appsail.dev/en/guide/ai-agent/) · [Hetzner alternatives](https://vps.appsail.dev/en/guide/hetzner-alternative/) · [Serving users in mainland China: Hong Kong and Japan](https://vps.appsail.dev/en/guide/hk-japan-vps-china/) · [VPS for Australia and New Zealand](https://vps.appsail.dev/en/guide/australia-vps/) · [Best VPS for Claude Code](https://vps.appsail.dev/en/guide/claude-code-vps/) · [Black Friday VPS deals 2026](https://vps.appsail.dev/en/guide/black-friday-vps-2026/) · [Stock changes (RSS)](https://vps.appsail.dev/feed.xml)

About the data: prices, data centres and bandwidth are as stated on each vendor's website. Each row is one plan: “Main server (8G)” is the cheapest 8G RAM plan we have verified for that vendor and route, for running Claude Code; “Relay (China-optimised route)” is the cheapest plan of any RAM size for that vendor and route, used as a proxy or relay. All prices are converted to a US dollar monthly price (¥6.68 = $1, mid-rate on 2026-11-17); where a yearly price exists we divide it by 12, and the monthly price is shown under “Price and discounts”. Route labels follow the vendor's own product names or descriptions; this site only traces the outbound route from China (the “Outbound route” column on vendor pages). The green tag “Native / Residential / Dual-ISP” after the route marks the IP type (of interest for cross-border e-commerce and overseas social media work), also as the vendor's product name or description states it; this site does not verify each IP. “CN2” means the vendor only says CN2, without stating GIA or GT; “All 3 carriers optimised” means the vendor states that return traffic for Telecom, Unicom and Mobile goes over optimised routes. The “Telecom xx · Unicom xx · Mobile xx ms” under the test IP is the average latency of pings from up to 12 public measurement probes on each of mainland China's three carriers; one vendor is measured each day at 5 a.m. Beijing time, in turn, so each vendor comes round about every ten days. The probes differ every run, so the figures are for reference only; hover to see the probe cities and loss rate. When one network shows 20% loss or more it is marked “Telecom 99 (loss 55%)”; such a route will lag in real use, and filters and sorting do not count it as reachable. IPs that block ping are tested on a TCP port instead. “no reply” on one network means that probe could not reach the IP; “Test IP unreachable” means probes both inside and outside China could not reach it, usually because the test IP has been retired; it does not mean the route or the server is unavailable. Hong Kong data centres are not in Anthropic's supported regions; other data centres are judged by the official list of supported countries, which does not mean every IP has been tested to work.

Some buy links carry an affiliate commission; commissions do not affect the ranking or the data in the table. Where a vendor has a public coupon code, it is shown in the “Price and discounts” column; enter it yourself at checkout.
