Plans & billing
Plan allowances: and what happens at the cap
Every plan meters exactly one number: the AI tokens its assistant spends answering in a calendar month. Nothing else is capped, and nothing dramatic happens at the limit: the assistant pauses politely and everything recovers on its own on the 1st.
The one thing a plan measures
| AI tokens / month | AI models read and write text in tokens; a page of text is roughly 500 of them. The allowance counts what the assistant reads and writes while answering your questions: the question, the document excerpts it consults, and the answer it writes. A typical answered question spends about 25,000 tokens. The counter resets on the 1st of each month (UTC). Starter carries 4M tokens, Team 20M, Enterprise 50M; the Unlimited tier has no allowance at all. |
|---|---|
| Re-reading is free | Context the assistant re-reads from its cache is not counted. Long conversations do not drain the allowance faster just because they carry history. |
| Documents, sources, people | Never limited. Connect everything, invite everyone; the size of your knowledge base is shown on your account page as information, and no plan caps it. |
Where you see the allowance
Three places, always in agreement: the Plan & usage page of your deployment's console, your account page at naxistechnologies.com, and the plan card you bought on the pricing page. Each shows the same live figure: tokens spent, tokens included, and the percentage used.
The 80% heads-up
When a deployment passes 80% of its monthly allowance, the administrator sees a notice on the account overview and receives one email from us for that month. Nothing changes at 80%; it exists so the cap never takes you by surprise.
What happens at 100%
| Answering pauses | The assistant answers every further question with a short note that the month's allowance is spent, without spending any more tokens. Nothing queues up, nothing errors, channels stay connected. |
|---|---|
| Everything else continues | Sources keep syncing, documents stay fresh, sign-ins work, the console works. Only answer generation waits. |
| Recovery is automatic | On the 1st the allowance resets and answering resumes by itself. An upgrade takes effect within minutes and resumes it immediately. There is never an overage bill; the assistant will not spend money you did not agree to. |
If a month runs hot
- Check who is asking what. The console's overview shows the most active people and topics; occasionally one integration or one experiment is responsible for most of a month's spend.
- Upgrade the plan. Plan & billing on your account page changes the plan in place: the new allowance reaches the deployment within minutes, no reinstall, nothing interrupted. The tiers are here.
Consistently beyond Enterprise? Write to us: the Unlimited tier is shaped per organisation and carries no allowance at all.