AI Service Center Beta

Manage every AI model and service on your NAS

Q: Is it safe to use Cloud Providers? Does my data leave the NAS?

If you use the Local Model Library , nothing leaves your NAS — the request and the response both stay local. If you use a Cloud Provider , the request is sent to that provider’s own service over the internet, using the API key you supplied. From that point on, your data is handled under that provider’s own privacy policy , not QNAP’s.

Q: Do I need to pay to use AI Service Center?

AI Service Center itself has no license fee . Using the Local Model Library costs nothing beyond the NAS you already have. Using Cloud Providers requires your own account with that provider. You pay the provider directly, based on your usage — AI Service Center does not add its own charge on top. You enter the provider’s unit prices yourself when you add it, so AI Service Center can estimate your spending.

AI Service Center is the AI management platform for your NAS. Local models and major cloud AI services are unified in a single console, where administrators control which models are available and set token and spending limits. Apps on the NAS obtain AI capabilities through one shared interface, with no per-app setup.

Download now Requirements

Why you need AI Service Center

Single console

Local and cloud AI services, managed together
Configure once

Every app shares the same set of models
Usage under control

Set limits on both token usage and spending

Overview

Configure once. Every app can use AI

Before: the problem

Every app connects on its own

Each app finds its own models and stores its own API keys. Overall usage is invisible, and cloud spending only surfaces when the bill arrives.

After AI Service Center

Models and keys, configured once

Models and keys are configured once, centrally. Apps simply declare the capabilities they need; AI Service Center decides which model to use and whether it runs locally or in the cloud, and only ever uses models the administrator has approved.

One shared interface

Apps

QuAgent

Qsirch

QuONE

Unified layer

AI Service Center Route · Control · Audit

Usage · Cost · Audit logs — every call is traceable

Where it runs

Local models

CPU · GPU · iGPU · NPU

Cloud services

OpenAI · Azure OpenAI · Gemini · Claude · OpenAI Compatible

Features

Four management pages covering models, cost, routing and audit

Select any page on the left to see its description and the actual product screen.

Model deployment without the command line

AI Service Center provides a QNAP-verified model list. Deployment takes two steps.

STEP 01

Pick a model from the list

Every model shows its file size, memory requirement, hardware compatibility, current status and the apps using it.

STEP 02

Choose the compute device

CPU, GPU, iGPU or NPU. Download and loading then complete automatically in the background, without interrupting other tasks.

Runtime packages

Installed per GPU environment

NVIDIA environments need NVIDIA AI Runtime and LLM Core; Intel environments need Intel AI Runtime and LLM Core (Intel).

Local Models screen in an NVIDIA environment

NVIDIA: a broader model selection, including the Gemma4, Qwen3.x and Nemotron families

Actual product screen

Cloud AI spending, visible as it happens

Cloud AI spending usually only becomes clear once the bill arrives. This page gives administrators an at-a-glance view of every provider’s status and spending — which are active, how much has been spent this month, and which one is approaching its budget limit.

Budget limits and alerts

Set a monthly budget and warning threshold as you add each provider; service stops automatically when the limit is reached.

Keys never leave the NAS

API keys are stored only on the local NAS and are never sent to QNAP servers.

Only enabled models are callable

Models the administrator has not enabled cannot be called by any app.

Spending, budget progress and provider connection status

Actual product screen

Every app-to-model mapping at a glance

Administrators assign a local model to each capability on a single page, decide whether cloud access is allowed, and choose which cloud models may be used. Apps register and declare their required capabilities automatically at install time.

Change once, applies everywhere

Switching a model takes one edit; every app picks it up.

Cloud access, app by app

Allow only specific apps to use cloud services and keep the rest local.

Only enabled models are listed

The list shows only models enabled in provider settings, preventing mis-selection.

Model assignment per app and per capability

Actual product screen

Audit without switching programs

System events related to AI Service Center are now in one console, with no need to open QuLog Center separately and switch between two interfaces.

Every call is recorded

The access log captures time, app, model, capability, token count, cost and latency.

Filter and export

Filter and sort by app, capability or status, and export to CSV.

Cost attributed per model

Cost is estimated from the prices you set and can be attributed to individual models.

Access log and system log

Actual product screen

Supported Apps

Configure once. Every app can use AI

Apps register with AI Service Center and declare their required capabilities at install time, so this list will keep growing. AI apps released later inherit the model and policy settings already in place.

QuAgent

The AI assistant on your NAS. Operate your NAS in natural language and make NAS management simpler than ever.

Learn more
Qsirch

The AI-powered search engine for your NAS. Semantic search finds files stored on the NAS in seconds.

Learn more
QuONE (Coming soon)

The NAS assistant in your pocket. Unified installation, AI intelligence, photo memories and device management make it the most intuitive entry point to your NAS.

Requirements

System requirements and hardware limits

Which AI Service Center features are available depends on your NAS processor architecture. Cloud Providers works on every supported model; Local Models requires an x86 model with a verified GPU.

Item	x86 models	ARM64 models
Supported features	Full functionality, including Cloud Providers and Local Models	Cloud Providers only; Local Models is not supported
Operating system	QTS 6.1.0 or later, or QuTS hero h6.1.0 or later	QTS 6.1.0 or later, or QuTS hero h6.1.0 or later
Cloud Providers	No GPU required	No GPU required
Connectable cloud services	Connect to OpenAI, Azure OpenAI, Gemini, Claude and OpenAI-compatible cloud services with an API key	Connect to OpenAI, Azure OpenAI, Gemini, Claude and OpenAI-compatible cloud services with an API key
Local Models	Requires a GPU that has passed compatibility verification, plus the matching runtime and model packages, so models load and run properly on the device	Not supported
Runtime packages	NVIDIA: NVIDIA AI Runtime, LLM Core Intel: Intel AI Runtime, LLM Core (Intel)	Not supported

AI SERVICE CENTER

Configure your models once.
Every app has AI instantly

Install AI Service Center from App Center to start configuring local models and cloud services.

Download now Requirements

Frequently asked questions

Local Model Library vs. Cloud Providers — which should I use?

Local Model Library runs the model on your NAS’s own hardware. Nothing leaves your NAS, and there is no per-use fee. You need a GPU or NPU that has passed compatibility verification, and you are limited to the models available for local use.
Cloud Providers send your request to an outside AI service using your own API key. You get access to larger or newer models without needing extra hardware, but each request goes out over the internet and you pay the provider directly.

Is it safe to use Cloud Providers? Does my data leave the NAS?

If you use the Local Model Library, nothing leaves your NAS — the request and the response both stay local.
If you use a Cloud Provider, the request is sent to that provider’s own service over the internet, using the API key you supplied. From that point on, your data is handled under that provider’s own privacy policy, not QNAP’s.

Do I need to pay to use AI Service Center?

AI Service Center itself has no license fee.
Using the Local Model Library costs nothing beyond the NAS you already have.
Using Cloud Providers requires your own account with that provider. You pay the provider directly, based on your usage — AI Service Center does not add its own charge on top. You enter the provider’s unit prices yourself when you add it, so AI Service Center can estimate your spending.

Why you need AI Service Center

Single console

Configure once

Usage under control

Configure once. Every app can use AI

Every app connects on its own

Models and keys, configured once

One shared interface

Local models

Cloud services

Four management pages covering models, cost, routing and audit

Model deployment without the command line

Pick a model from the list

Choose the compute device

Installed per GPU environment

Cloud AI spending, visible as it happens

Budget limits and alerts

Keys never leave the NAS

Only enabled models are callable

Every app-to-model mapping at a glance

Change once, applies everywhere

Cloud access, app by app

Only enabled models are listed

Audit without switching programs

Every call is recorded

Filter and export

Cost attributed per model

Configure once. Every app can use AI

QuAgent

Qsirch

QuONE (Coming soon)

System requirements and hardware limits

Configure your models once.Every app has AI instantly

Frequently asked questions

Configure your models once.
Every app has AI instantly