ReadyServer includes AI model access with every VPS plan, without per-token billing
- Written by: Telegraph Magazine
The OpenAI-compatible API lets developers connect applications running on their VPS to ReadyServer's shared AI inference infrastructure, without arranging a separate model service, obtaining an API key or paying per token.
For a developer building an AI application, hosting the application and accessing the model behind it are often separate decisions. A team may already have a server for its code and data, but still needs to work out where the model will run, how to connect to it and how usage will be billed. ReadyServer's approach brings model access into its existing VPS offering.
"Developers may have a clear idea for an AI application, but getting access to a model introduces another service to set up and another usage bill to manage," said Alan Woo, Chief Executive Officer of ReadyServer. "We wanted to make that first step simpler by including model access with the VPS they already use."
The API is reachable from a ReadyServer VPS through the company's private network. It is not exposed for use from a customer's own computer or a server hosted with another provider. Customers do not receive a dedicated GPU VPS or need to install a model on their existing server. Instead, applications on their VPS send requests to models hosted on ReadyServer's shared inference infrastructure.
Developers can use the service to build chatbots, AI agents, coding assistants and document search tools. It also supports retrieval-augmented generation (RAG), which lets an application draw on selected documents when producing a response. Other supported capabilities include summarisation, classification, data extraction, image input, tool calling and structured JavaScript Object Notation (JSON) output. These are available capabilities; the applications a customer builds will depend on its own development work.
The API uses a format compatible with the OpenAI API. Developers can connect using tools such as curl or the Python and Node.js OpenAI Software Development Kits (SDKs) by configuring their applications to use the ReadyServer API endpoint. No additional service installation is required on the VPS. While some SDKs require a non-empty API key field, ReadyServer does not issue an actual key for this service and a placeholder value can be used.
ReadyServer currently serves two open-weight Qwen models. Qwen3.6-27B is designed to return direct answers more quickly. Qwen3.8-27B has a longer context window and is suited to work involving lengthy documents or tasks that require multiple steps. Both are hosted by ReadyServer; "OpenAI-compatible" describes the API format, rather than the models themselves.
"We chose shared inference capacity so customers can access hosted models without each having to run a dedicated GPU server," Woo said. "It is a practical way for developers to start building, but shared capacity also means we need to be clear about fair use and what happens when demand is high."
The service has no published monthly token allowance and does not charge customers per token. Usage is governed by ReadyServer's fair-use policy. When the service is busy, requests may be queued and response times may increase. The ReadyServer LLM API is provided on a best-effort basis without a service-level agreement.
The ReadyServer LLM API is available to customers across every ReadyServer VPS plan. Documentation covering setup, supported models, API use and fair use is available at readyserver.ai.
Hashtag: #ReadyServer
The issuer is solely responsible for the content of this announcement.
About ReadyServer
ReadyServer Pte Ltd provides internet infrastructure, network, data centre and cloud computing solutions. The ReadyServer LLM API gives its VPS customers access to hosted open AI models through the company’s private network.
