A prepaid balance in euros: you pay for what you use, not for a package
In AraLith COS, AI is billed on usage, exactly like the model APIs it runs on. There are no credits to count and no packages to choose: you top up an amount in euros and every generation draws down the real cost of the tokens it used, input and output, at the model's official rate. The balance never expires, and the cost of each answer is shown right underneath it.
The AI balance is a prepaid amount in euros attached to your AraLith COS environment. It is not an internal currency with an exchange rate to learn: it is money, and every operation the assistant performs draws down its actual cost.
The cost of an answer is calculated on the tokens that answer consumed — input, cache reads and writes, and output — at the official rate of the model used. Tools that call external services have a price per run instead. Every other tool only costs the tokens it generates.
The detail of every conversation is on the Usage page in the panel: by conversation, agent profile, model, tool and day, with CSV export.
There is no price list per operation: the cost depends on how much text actually goes in and comes out. This table says what you pay for in each case.
| Operation | What you pay for |
|---|---|
| Chat question, no tools | Only the request and answer tokens |
| Reading a page or searching content | The tokens, plus the content read |
| SEO audit of a page | The tokens of the analysis |
| Keyword research or SERP analysis | The tokens, plus the service's per-run price |
| Creating a page or an article | The whole sequence: longer page, higher cost |
| Translating content | The tokens of source text and translation |
| Generating images or video | The per-run price from the media rate card |
| A full SEO pipeline | All its tokens, plus any external services called |
The exact cost appears under every answer and on the Usage page, broken down by conversation, model and day.
Four factors explain almost the entire difference between an answer that costs a few cents and one that costs a great deal more. Knowing them is the practical way to keep spending where you want it.
official rate per million tokens
Different models have different prices. Simpler agent profiles can be routed to a lighter model, and the saving shows up on the Usage page.
input and output tokens
The more context goes into the request and the more text comes out in the answer, the more that answer costs. It is the factor your own way of working affects most.
repeated input costs a fraction
The stable part of the prompt is served from cache at a fraction of the price. A high cache-read to input ratio means you are paying very little for recurring context.
priced per run
SERP analysis, competitor research, keyword research, AI visibility checks, images and video are priced per run. Every other tool only costs the tokens it generates.
the Usage and balance menu item
In your AraLith COS admin area you will find the available balance and the Top up balance button. The same balance appears in the header of the chat widget: clicking it brings you here.
quick amount or your own
Pick €10, €25, €50 or €100, or enter your own amount in whole euros with a €10 minimum. No package to weigh up: you decide how much to load.
card or PayPal, electronic invoice
Complete the payment by card or PayPal and the balance is available immediately. The electronic invoice is issued to the billing details you provide.
warned before it runs out
The Usage page shows costs by conversation, agent profile, model and day, and exports to CSV. Set a low-balance threshold and you get the warning before the balance runs out.
the Usage and balance menu item
In your AraLith COS admin area you will find the available balance and the Top up balance button. The same balance appears in the header of the chat widget: clicking it brings you here.
quick amount or your own
Pick €10, €25, €50 or €100, or enter your own amount in whole euros with a €10 minimum. No package to weigh up: you decide how much to load.
card or PayPal, electronic invoice
Complete the payment by card or PayPal and the balance is available immediately. The electronic invoice is issued to the billing details you provide.
warned before it runs out
The Usage page shows costs by conversation, agent profile, model and day, and exports to CSV. Set a low-balance threshold and you get the warning before the balance runs out.
No. The balance you top up stays available until you use it, with no time limit. Top-ups are non-refundable, except where the law requires otherwise.
Ten euros. You can pick a quick amount of €10, €25, €50 or €100, or enter your own amount in whole euros, always from €10 upwards. Top up whenever you like, without changing plan and without waiting for renewal.
The system checks the balance before every request. With the balance empty the request does not start, the chat shows a warning and the conversation closes: nothing is charged automatically and no renewal is triggered behind your back. Every non-AI feature of the CMS keeps working normally. Once you top up, you carry on from where you left off.
The cost appears under every single answer from the assistant. On the Usage page you find it broken down by conversation, agent profile, model, external tool and day, with input and output tokens, the rate applied and the amount billed, all exportable to CSV. In the wizards, before you start, you see the historical estimate of what that operation usually costs.
Yes, in three concrete ways. The prompt cache makes the recurring part of the context cost a fraction: if the ratio of cache-read tokens to input tokens is high, you are already saving. From the AI settings you can route simpler agent profiles to a cheaper model and keep dynamic routing on, so light requests do not pay for the large model. And the global rules in the SEO project spec, which enter the prompt on every turn, are worth keeping short.
Choose your plan, top up the AI balance when you need it, and pay only for what you use.
Content created with the support of artificial intelligence.
