Install and configure 9router on a VPS
An AI API router, not a network router
VPS-1
From
S$5.78
ex. GST/month
S$6.30 incl. GST/month
Specifications:
2 vCores
4 GB RAM
40 GB SSD NVMe
Daily backup of the previous 24 hours
VPS-2
From
S$10.88
ex. GST/month
S$11.86 incl. GST/month
Specifications:
4 vCores
8 GB RAM
75 GB SSD NVMe
Daily backup of the previous 24 hours
VPS-3
From
S$15.64
ex. GST/month
S$17.05 incl. GST/month
Specifications:
6 vCores
12 GB RAM
100 GB SSD NVMe
Daily backup of the previous 24 hours
VPS-4
From
S$30.00
ex. GST/month
S$32.70 incl. GST/month
Specifications:
8 vCores
24 GB RAM
200 GB SSD NVMe
Daily backup of the previous 24 hours
Why host 9router on a VPS
A VPS provides an isolated, configurable environment available online to operate 9router VPS without depending on a local machine. You control the application path, system rules, and open services, while keeping the flexibility needed to modify your configuration. NVMe SSD storage helps to process logs, configuration files, and service states quickly. The VPS network supports API exchanges with latency suitable for interactive use cases. Depending on your technical constraints, you can also enable Caveman mode, adjust it, or disable it, within a legal usage framework defined by your organisation.
Manage your LLM calls from a single point
An entry layer for your AI applications
Once installed, 9router becomes the entry layer for your AI applications. Your service no longer calls each LLM provider directly. It sends a request to 9router, which applies your routing rules before forwarding the request to the chosen model. You can thus organise calls according to the type of task, expected cost, availability or your team's preferences.
A common interface for your integrations
This architecture simplifies contact between your applications and the LLM ecosystem. Instead of maintaining several scattered integrations, you keep a common interface. Changes of provider, parameter adjustments or strategy evolutions are managed on the software router side, without imposing a redesign on every connected application.
A global VPS infrastructure
The OVHcloud VPS infrastructure relies on global data centres and internet access designed for online services. This foundation helps to control latency, maintain continuity of access and bring your services closer to the areas where your users interact with your tools.
Reduced technical dependency
Routing also reduces dependency on a single technical provider. If a model becomes less suitable for a use case, you can direct certain requests elsewhere, without changing the entry point on the application side. This point is important for teams building agents, assistants or business services that are set to evolve regularly.
Exploit 9router with a solid operational base
Reduce token consumption and manage your quotas
Request optimisation
The cost of an AI application does not only depend on the number of users. It also depends on the length of the requests, the expected responses and repeated calls to the models. RTK Token Saver acts on this precise point: it optimises or compresses requests before they are sent, in order to reduce the tokens consumed when the context allows it.
API quota management
This control becomes useful as soon as several teams share the same public API or several keys. API quotas provide a framework: they help to avoid overruns, smooth out activity and identify usage that consumes more than expected. You can thus distinguish a highly solicited internal agent, a poorly configured business tool or a spike coming from your community networks.
Varied use cases
The use cases are varied. A support centre can reduce redundant prompts in assisted responses. A product team can manage the calls of an assistant integrated into its software. A community can offer AI functions without letting each service consume without limit. Even in games or interactive experiences, continuous optimisation prevents frequent exchanges from degrading budgetary control.
Documentation of the rules
To follow these practices, your team can also document its rules, share an internal technical newsletter and adjust policies according to the observed usage feedback.