Hosting since 2011
4.5/5 on Trustpilot

AnythingLLM VPSYour documents, your AI agents, on your own server in France

  • AnythingLLM deployed automatically
  • Chat with your documents, with cited sources
  • AI agents, chat widget and API
  • Hosted in France, 24/7 support
Deployed for you
AnythingLLM ready after ordering
Your model
Ollama or a provider's API
Automatic HTTPS
Generated address or your domain
24/7 support
From real people

Pick your AnythingLLM VPS

AnythingLLM is deployed automatically on every plan on this page, all above the chosen minimum: 2 vCPU, 4 GB of RAM and 30 GB of disk.

Operating system
  • Ubuntu26.04 · 24.04 · 22.04
  • Debian13 · 12 · 11
Pick one when you order

VPS delivered in 10 seconds

  • HP-4
    Processor2 vCPUAMD EPYC 9554
    Memory4 GBDDR5
    Storage40 GB NVMeor 80 GB SSD
    Network1 Gbps
    €6.39excl. VAT/month
    €7.99Save 19€/year
    Choose
  • HP-8
    Processor4 vCPUAMD EPYC 9554
    Memory8 GBDDR5
    Storage80 GB NVMeor 160 GB SSD
    Network1 Gbps
    €12.79excl. VAT/month
    €15.99Save 38€/year
    Choose
  • HP-16Most popular
    Processor8 vCPUAMD EPYC 9554
    Memory16 GBDDR5
    Storage160 GB NVMeor 320 GB SSD
    Network1 Gbps
    €25.59excl. VAT/month
    €31.99Save 77€/year
    Choose
  • HP-32
    Processor16 vCPUAMD EPYC 9554
    Memory32 GBDDR5
    Storage320 GB NVMeor 640 GB SSD
    Network1 Gbps
    €51.19excl. VAT/month
    €63.99Save 154€/year
    Choose
  • HP-64
    Processor32 vCPUAMD EPYC 9554
    Memory64 GBDDR5
    Storage640 GB NVMeor 1280 GB SSD
    Network1 Gbps
    €102.39excl. VAT/month
    €127.99Save 307€/year
    Choose

Every plan has everything you need, and more

  • AnythingLLM deployed
  • Automatic HTTPS
  • Unlimited traffic
  • DDoS protection
  • 24/7 French-speaking support

Why AnythingLLM on an OnetSolutions VPS

Recent processors, NVMe storage and protected data, under every AnythingLLM server.

Intel Xeon or AMD EPYC, NVMe storage

Intel Xeon with DDR4 on the General Purpose range, AMD EPYC 7443 on the Performance range, AMD EPYC 9554 with DDR5 on the High performance range.

Up to 1 Gbps

Up to 300 Mbps on the General Purpose range, 1 Gbit/s on the Performance and High performance ranges, unlimited traffic everywhere.

OnetSolutions API

Automate deployments, manage your servers and connect your AI assistants to your infrastructure. See the documentation

Backups and snapshots

AnythingLLM volumes are backed up every day on the server. Daily VPS backups available as an option from €1.99 excl. VAT/month, plus a free first snapshot.

Your document assistant in three steps

Nothing to install: you order, AnythingLLM is deployed on your server with an HTTPS address.

  1. 01

    Pick your VPS

    Choose a plan: AnythingLLM is already selected as a One Click App in the order.

  2. 02

    AnythingLLM deploys

    Automatically once the server is created, with a generated HTTPS address and an access password generated at deployment.

  3. 03

    Connect a model

    Choose the model provider (Ollama or an API), create a workspace and drop in your documents.

AnythingLLM, the all-in-one AI application

Documents, agents and users in a single application. The language model comes from the provider you connect to it.

  • Workspaces

    Each workspace has its own documents and conversations, and can use its own model.

  • Chat with your documents

    Drop in PDF, TXT, DOCX and other formats: answers cite the passages used. A query mode answers only from your documents.

  • AI agents

    With @agent, the model can search the web, scrape a page, list and summarise your documents or produce a chart.

  • No-code Agent Flows

    Build agent skills in a visual editor, without writing code.

  • MCP compatible

    Add Model Context Protocol servers: their tools become available to your agents.

  • Multiple users

    Multi-user mode with admin, manager and default roles, and access set per workspace.

  • Chat widget for your site

    Publish a workspace as an embeddable chat on your website.

  • Developer API

    A full API, documented on your instance at /api/docs, to manage workspaces, add documents and chat.

  • Built-in vector database

    LanceDB runs inside the application: your document text and embeddings stay on your server, with no configuration.

No model included: you choose yours

The server version of AnythingLLM does not ship a language model. Connect an Ollama server you run, or a provider's API with your own key, and switch whenever you like.

  • Ollama
  • LM Studio
  • OpenAI
  • Anthropic
  • Mistral
  • Google Gemini
  • Groq
  • OpenRouter
  • Azure OpenAI
  • AWS Bedrock

Embeddings, on the other hand, can be computed on the server: AnythingLLM provides a built-in embedding model (all-MiniLM-L6-v2, trained mostly on English) that runs on the processor. Leaving it 2 GB of memory, as its documentation recommends, is why this page starts at 4 GB of RAM.

What about a GPU?

AnythingLLM does not use a GPU itself: its built-in embedding engine runs on the processor. To run local models faster, the NVIDIA L4 GPU option (24 GB of VRAM, €0.70 excl. VAT/hour billed by the minute, no commitment) can be added to High performance VPS: it benefits an Ollama server you install there, which you then connect to AnythingLLM.

AnythingLLM is open source under the MIT licence, published by Mintplex Labs. AnythingLLM's anonymous telemetry is disabled by default in this deployment.

What you can do with AnythingLLM

An assistant that knows your documents, on your server, with the model you choose.

  • Internal knowledge base

    Procedures, contracts, product documentation: each team queries its documents and sees the sources behind the answers.

  • Chat on your website

    A chat widget powered by a workspace answers your visitors from your documentation.

  • Research agents

    Ask an agent to read a web page, summarise a document or prepare a chart.

  • The model of your choice

    A local model through Ollama for sensitive documents, an API for the rest, workspace by workspace.

    • Ollama
    • Anthropic
    • Mistral
  • Tools through MCP

    Plug in MCP servers to give your agents access to your own tools and data.

    • MCP
    • tools
    • data
  • API integration

    Feed your workspaces and query them from your applications with the developer API.

AnythingLLM runs on your VPS, our console runs the server and the application

Your VPS and AnythingLLM are managed from the OnetSolutions console, without SSH, and you keep root access to the server.

  • AnythingLLM status, logs and environment variables
  • Your domain, with an automatic HTTPS certificate
  • Updates detected and applied from the console
  • Daily backup of the application volumes, 7-day retention
console.onetsolutions.net/compute/instances/anythingllm-prod
OnetSolutions
Agence Horizon
Compute/Instances/web-prod-01
web-prod-01Running
GP-8 · Intel Xeon 2,60 GHz
RestartConsole
CPU34%
RAM5,2 / 8 Go
Network128 Mb/s
System
Ubuntu 26.04 LTS
IP address
51.210.24.8
Uptime
34 j
Size
4 vCPU · 8 Go
CPU activity (24h)34%
-24 h-12 hnow

A team behind your server

Technicians who know your VPS, reachable at any hour.

4.5/5 on Trustpilot
Contact support

24/7 French-speaking support

Open a ticket from the console, day or night: a person answers you about your server.

Help center

One Click Apps guides: deployment, domains, backups and updates.

30-day money-back guarantee

Try AnythingLLM on your VPS: if it doesn't suit you, you get refunded.

MCP · Beta

Manage your VPS with AI

Our MCP server connects Claude, Cursor or any Model Context Protocol client to your OnetSolutions account: ask about your servers and their status in natural language, without leaving your assistant.

  • Read-only: the assistant looks, it changes nothing
  • Signs in with your OnetSolutions account
Set up the MCP server
  • Claude
  • Cursor
  • Any MCP client

AnythingLLM FAQ

What to know before you order, and right after.

What is AnythingLLM?

AnythingLLM is an all-in-one AI application published by Mintplex Labs: workspaces, chatting with your documents (RAG), AI agents, user management, a chat widget and an API. This One Click App deploys its server version on your VPS.

Is an AI model included? Do I need a GPU?

No language model is included: the server version of AnythingLLM does not ship one. You connect an Ollama server you run or a provider's API (OpenAI, Anthropic, Mistral, Groq, OpenRouter…) with your own key. AnythingLLM itself does not need a GPU. If you want to run local models faster, the NVIDIA L4 GPU option (€0.70 excl. VAT/hour, billed by the minute) can be added to High performance plans, for an Ollama server installed on that VPS.

Where are my documents stored?

On your VPS, in a datacenter in France: documents, conversations and the LanceDB vector database. The text of the relevant passages is sent to the model to generate the answer: with an API it is processed by its provider; with an Ollama server you host, it stays on your infrastructure.

How do I access AnythingLLM after ordering?

Once the VPS is created, AnythingLLM is deployed automatically. Open the application in the One Click Apps section of the console and follow its HTTPS address. The instance is protected by a password generated at deployment: find it among the credentials in the Overview tab or in the environment variables (AUTH_TOKEN). Then add your domain in the Domains tab: the HTTPS certificate is issued automatically.

Can my whole team use it?

Yes, by turning on multi-user mode in the AnythingLLM settings: you create the administrator account, then your team's accounts with the admin, manager or default role. Note that this mode cannot be turned off once enabled.

What minimum configuration do I need?

AnythingLLM's documentation recommends 2 GB of RAM and a 2-core processor, plus 2 GB of available memory for its built-in embedding engine. One Click Apps reserve 1 GB for the system and the platform: this page therefore starts at 2 vCPU, 4 GB of RAM and 30 GB of disk. You can move to a larger size from the console.

Is my data backed up?

AnythingLLM volumes (documents, vector database, settings) are backed up every day at 3 a.m., with 7 days of retention, on the VPS disk: they protect you from a mistake in the application, not from losing the server. For that, add the VPS backup option (from €1.99 excl. VAT/month) or take a snapshot.

How do updates work?

Every night, the console checks whether a new AnythingLLM version has been published and shows it on the application page. Nothing is updated without your action: one click on "Update" is enough, and the data volumes are kept.

Who maintains AnythingLLM?

AnythingLLM is developed by Mintplex Labs and its community. OnetSolutions provides the server, the deployment and the console tools; configuring the application and your models is up to you. Our support covers the VPS: server, network, storage and availability.

Is AnythingLLM paid?

No. AnythingLLM is open source under the MIT licence. The prices on this page are those of the VPS; model API usage is billed by its provider.

Launch your document assistant

Pick your plan: the server arrives with AnythingLLM deployed, ready for your documents and your model.

HP-4AnythingLLM included
from€6.39excl. VAT/month

€7.99with the annual commitment

  • 2 vCPU AMD EPYC 9554
  • 4 GB DDR5
  • 40 GB NVMe
  • 1 Gbps
Order HP-4
AnythingLLM VPS from €6.39 excl. VAT/month, hosted in France - OnetSolutions