Skip to content

How much will the AI cost me?

Your agent runs on your LusoVPS server, but the "thinking" is done by an AI model from a provider such as Anthropic (Claude), OpenAI or OpenRouter. This page explains how that AI is billed, what makes the bill go up, and how to keep it low.
Time needed 5 minutes to read
You need Nothing yet. Later: an account and an API key with an AI provider
Result You know what you pay for, and how to keep the AI bill under control

Two separate bills

Bill Who sends it What it pays for
Your plan (Lite, Plus or Extra) LusoVPS The server and the agent installed on it. See invoices and payments.
The AI Your AI provider What your agent uses of the AI model.

You create an account with the AI provider, get an API key and paste it into your agent. The provider bills you directly. LusoVPS adds no markup.

How the provider counts your use

Providers bill per use, not a fixed monthly price for the agent. They measure use in tokens: small pieces of text. Every message your agent sends to the model, and every answer it gets back, uses tokens.

You can see tokens in some places:

  • n8n: after a test chat, Logs shows the time and the tokens used. See n8n: sign in and build your first AI agent.
  • OpenClaw: Pages → Usage shows the tokens and what they cost, by model and by provider. That cost is calculated by OpenClaw: the real bill comes from your provider. See A tour of the OpenClaw dashboard.
  • Your provider's console: the exact amount you will pay.

We do not list provider prices here: they change, and they differ by model. Your provider's price page shows the cost of each model.

What makes the cost go up

  • Bigger models. Bigger models are smarter but cost more per message. Smaller ones are faster and cheaper.
  • Long conversations. The model reads the conversation so far each time it answers, so a long conversation uses more tokens per message than a new one. Hermes keeps its memory short for the same reason: memory is loaded into every conversation.
  • Frequent scheduled tasks. Each run of a scheduled task uses your AI provider, like a normal conversation. A short job once a day is cheap; a job that runs every few minutes is not. The same goes for n8n workflows with an AI step that run on a schedule. (Simple checks that Hermes sets up as script-only jobs, like the website alert, run without the AI.)
  • A leaked API key. Anyone with your key can use your credit. Treat it like a password.

How to keep the cost down

Use a smaller model for simple tasks

For everyday questions, replies and summaries, a smaller model (for example Claude Haiku) is usually enough and costs less. Keep the bigger models for long or difficult tasks.

Start a new chat for a new topic

In the Hermes dashboard Chat, click New chat. In OpenClaw, click New conversation (the red +). Hermes still remembers the facts in its memory.

Choose a sensible schedule

  • Hermes: run jobs only as often as you need them. Use Pause on the CRON page for jobs you do not need for now. See Schedule a recurring task.
  • n8n: check Trigger Interval in the Schedule Trigger. Minutes with 1 runs every minute. Unpublish test workflows when you are done. See Run a workflow on a schedule.

Set a spending limit with your provider

AI providers such as Anthropic let you control your spending in their console, with prepaid credit or a spending limit. Set a limit you are comfortable with.

When the credit runs out or the limit is reached, the agent stops answering (or answers with a credit or billing error) until you add credit or raise the limit. See My agent stopped answering.

Keep the key safe

Do not paste your API key in chats, emails or screenshots. If it leaks, delete it in your provider's console, create a new one and paste it in your agent. See Connect Hermes to Claude.

Common problems

The agent answers with a credit or billing error

Your provider account has no credit left, or reached its spending limit. Add credit or raise the limit in your provider's console.

Does LusoVPS include AI credit in the plan?

No. Every plan uses your own AI keys. You pay the AI provider directly for what you use, with no markup from LusoVPS.

Do n8n templates cost money?

Copying a template is free. Some templates connect to paid services (for example an AI provider): those services bill you directly. See n8n: start from a template.

Next steps