AI Agent Hosting

Built for AI agents that do useful work

Give your agent the computing resources to turn requests into working services. Choose a plan for the workload you want to run, with resources and support agreed before activation.

• Run tasks between conversations. Host your agent runtime, scheduled jobs and integrations without keeping a personal computer switched on.
• Keep working data between sessions. Store your application’s task history, memory database, files and outputs on persistent storage.
• Connect tools and deliver services. Run APIs, web applications and automation that turn your agent’s capabilities into useful services for customers.
• Choose how you use models. Use a CPU plan for agents connected to model APIs, or dedicated GPU VRAM to run compatible models in your own hosting environment.
• Grow with your workload. Ask us about a larger plan as your resource needs increase. NetizenWorks is your point of contact for the agreed infrastructure and support.

CPU plans and prepaid pricing

Save 10% with one year prepaid, or 20% with two years prepaid. The totals below cover the full selected term.

* Our monthly prices are reference rates; billing is for 6, 12 or 24 months in advance. Discounts apply to the base hosting price only.

CPU resources included

CPU plans A–G provide Linux virtual servers with shared CPUs and SSD storage. The RAM below is system memory, not GPU VRAM.

• CPU: Regular shared vCPUs. CPU capacity is shared with other virtual machines; the physical processor model can vary.
• GPU and VRAM: CPU plans A–G do not include a GPU or dedicated GPU memory.
• Operating system and access: Linux; the distribution, SSH access and setup scope are confirmed before activation.
• Networking: One public IPv4 address is included. Incoming data transfer is free; included outgoing transfer is shown per month above.
• Storage and region: Local SSD capacity is shown above. Region availability is confirmed when we quote; backups and additional storage are optional extras.

Transfer allowances reset monthly. Extra outbound transfer and additional resources are chargeable; rates are confirmed in your quotation.

GPU / VRAM plans for AI

Run compatible language models, image generation and other GPU workloads in a Linux GPU environment arranged through NetizenWorks. Each plan includes one NVIDIA GPU, 100 GB of persistent network storage and a 20 GB temporary container disk.

* Monthly reference prices cover one GPU instance running throughout the month and the storage listed above. Six-month minimum; the totals cover 6, 12 or 24 months paid in advance. GPU and region availability are confirmed in your quotation before payment.

GPU resources included

* Listed configurations; system RAM and vCPU allocation can vary by host. We confirm the exact resources and region in your quotation. Model compatibility and capacity depend on precision, context length and concurrent requests.

• Persistent storage: 100 GB standard network volume for model files, data and saved work. It is separate from the GPU instance and retained when the instance stops or is replaced while the storage service remains active. Backups are a separate service.
• Temporary storage: 20 GB container disk for the environment and temporary files. This disk is cleared when the instance stops or is deleted; keep work that must survive on the persistent volume.
• Environment and access: Linux container with SSH access and agreed HTTP/TCP services. Access is to the GPU container; host administration, nested Docker and UDP services are unavailable.
• Networking: No per-GB inbound or outbound transfer charge for these GPU plans. Network speed and exposed ports depend on the selected host; a dedicated static public IP is not included.
• Configuration: On-demand GPU hosting, subject to availability. We confirm the template, model setup and support scope before activation. Additional storage and setup work are quoted separately.

Choose resources for your workload

Use the server for agent runtimes, scheduled tasks, integrations, websites or applications. The 8 GiB and 16 GiB plans provide more memory and CPU capacity for tools, browser automation, databases and multiple services; actual capacity depends on your workload.

Plan A has limited memory, so tell us what you intend to run before choosing it. We will confirm a suitable plan, available region, access arrangements, setup and support scope with you before activation.

Does an AI agent need a GPU?

For an agent that uses a model API, a CPU server can be enough. The server runs the agent application, tools and schedules, and stores its files and task history. The external model provider runs the language model. Model API usage is billed separately.

Running the language model on your own server is a different workload. A GPU with sufficient VRAM generally makes model inference much faster. The memory needed depends on the model, its precision, context length and number of simultaneous requests. Smaller models can also run in system RAM on a CPU, with lower performance; suitability needs benchmarking.

Choose a CPU plan when your agent uses an external model API, or a VRAM plan when you need a GPU to run models. System RAM and GPU VRAM are separate resources. Tell us whether you will use a model API or run a local model so we can assess the right resources.

Before you start

• One point of contact. NetizenWorks arranges the server and bills you for the agreed service.
• Extras are separate. Model API fees, optional backups, extra storage, transfer overages and applicable taxes are outside the listed base prices.
• Confirm the details first. Availability, activation, renewal, cancellation, backup responsibilities and any additional work are confirmed in a written quotation and service agreement before payment.
• Authorized use. Customers and agents acting on their behalf must use the service lawfully and follow the infrastructure provider’s terms and our AI Use & Commerce notice.

Payment options for agents and their customers

We accept PayPal, card payments through PayPal, bank transfer and cryptocurrency. Choose a method you are authorized to use, or arrange payment through your customer or sponsor.

• PayPal and cards via PayPal: Pay with your PayPal account or a card through PayPal. Request a payment link for your selected plan and prepaid term. For these payments, we apply a fraud review period of 3 business days after receiving payment, before hosting activation.
• Bank transfer: Use the bank details and payment reference provided with your invoice.
• Cryptocurrency: Tell us the token and network your wallet supports. We confirm a supported option, the amount, receiving address and any required memo before you transfer.

Prices are quoted in USD. Any currency conversion, payment fees and crypto quote expiry are confirmed with the payment instructions. Hosting is activated after payment is confirmed, any applicable fraud review period is complete, and the setup arrangements are agreed.

Email [email protected] for your invoice and payment instructions. For crypto payments, use only the token, network and destination specified for that invoice.