What are tokens and why you pay for them separately
This is the question we get most often. “If I pay a subscription, why do I pay for something else separately?” The answer is simple and - as it turns out - very much in your favour.
A token is a “piece of text”
AI models do not read letters or words the way we do. They split text into small pieces - tokens. Imagine building blocks: one short word is usually one block, a longer one splits into two or three.
Every question you ask and every answer the assistant gives is a number of such blocks. For each portion the AI provider charges a fee. That is the whole secret of the “fuel” we talk about.
Why we do not include it in the subscription
Because it would be unfair. AI usage is individual: one company sends a short question twice a day, another orders an analysis of hundreds of invoices and writes long documents. If we hid the fuel cost in the subscription, light usage would pay for heavy usage.
So we separate two things:
- Subscription - a fixed amount for the program, updates and support. It never grows.
- Tokens - the fuel cost, paid directly to the AI provider, exactly for what you use. It does not pass through us, we add no margin.
How much does it really cost
More than you think? Rather less. By default the assistant runs on the model DeepSeek v4.1 Flash - chosen precisely because it is extremely cheap for everyday office work.
| Situation | Model | Usage |
|---|---|---|
| Typical office company | DeepSeek v4.1 Flash (default) | max about $10 / week |
| Heaviest possible case | DeepSeek v4.1 Flash (default) | upper limit about $50 / week |
And on top of that, the assistant has a usage counter and a spending limit you set yourself. When the limit is reached, the program stops - so you never see a bill you did not expect.
Premium models - approach them consciously
The assistant also offers the strongest models (Claude Fable, Opus, Sonnet). They are brilliant but genuinely expensive - one question can “burn” 2 to 4 zl. That is why they are clearly marked and described, so nobody turns them on by mistake. For everyday work the default model is more than enough.
Short summary
Subscription - fixed, predictable.
Tokens - you pay for what you really use, no margin.
Limit - you set it yourself and never exceed it.
Result: full control over spending and no surprises.
Check what it would really cost you - on the free month.
The first month is 100% free. No card, no commitment.