Product Terms
This document covers the commercial and operational terms specific to ASAI: which plans exist, how usage is measured, how the API works, and what happens if an account is suspended.
It supplements the company-wide Terms of Service. Where this document is silent, the Terms apply. Where this document is more specific, this document applies.
Effective: 31 July 2026 · Version: 2.0
Operator: APERTURESyndicate OÜ (Republic of Estonia (EU), Registry code 17384111)
1. Plans
ASAI has three plans. Prices below are the standard EU prices in EUR; tax is applied at checkout according to your country.
1.1 Free — €0
- The smaller model tier
- A daily chat allowance, a weekly roleplay allowance, and a monthly image allowance (Section 2)
- No Code mode
- A small number of private characters and worlds
- Cannot publish characters or worlds to the public library
1.2 Pro — €18/month or €190/year
- The larger model tier
- Substantially higher chat, roleplay and image allowances, plus a daily Code allowance
- More private characters and worlds
- Can publish characters and worlds to the public library (see Creator Program)
- API access (Section 4)
1.3 Etherane — €90/month or €900/year
- Everything in Pro, with the highest allowances on every pool
- Unlimited private characters and worlds
- Dataset and model-training tooling
1.4 Platform bundles
The APERTURESyndicate Pro bundle (€30/month) includes ASAI at the Pro level alongside other platform products. The Supporter and Standard bundles do not include ASAI. See the Subscriptions page.
1.5 Roleplay add-ons
Two add-ons stack on any plan, including Free. They are mutually exclusive — you can hold one, not both.
- Character · Spark — €5/month. One dedicated ASAI Character with persistent in-character memory, a custom persona, and a private roleplay channel.
- Character · Cast — €10/month. Up to five such Characters, each with their own memory and world, plus session export.
Add-ons expand what you can build. They do not change the usage allowances in Section 2.
1.6 What does not exist
For the avoidance of doubt, and because an earlier version of this document described them:
- There is no pay-as-you-go option. You cannot buy extra hours. When an allowance is exhausted, it is exhausted until it resets.
- There is no Creator Credits system. Publishing characters does not earn you compute, credits, points, or money of any kind. See Creator Program.
2. How usage is measured
2.1 Tokens, presented as hours
Everything the model reads and writes is counted in tokens. For display we convert tokens into AI-hours at a fixed rate of 100,000 tokens = 1 AI-hour, because "2.3 hours left" is easier to reason about than a seven-digit token count.
An AI-hour is therefore a unit of work, not a unit of clock time. It does not correspond to how long you sat at the keyboard, and it is not the "seconds of compute × 2" formula an earlier version of this document described. Reading old messages, scrolling, and editing characters cost nothing — only generation is counted.
2.2 Task multipliers
Different kinds of work cost different amounts per token, reflecting what they cost us:
| Task | Multiplier |
|---|---|
| Chat (text) | 1.0× |
| Roleplay | 1.2× |
| Code | 1.5× |
| Messages containing images (vision) | 1.5× |
| Image generation | 2.5× |
2.3 Four independent pools
Your plan does not give you one allowance. It gives you four, and they do not borrow from one another. Running out of roleplay does not affect your chat allowance, and vice versa.
| Pool | Covers | Resets |
|---|---|---|
| Chat | ordinary assistant conversation | daily, 00:00 UTC |
| Roleplay | character and world conversation | weekly, at the start of the ISO week (Monday, 00:00 UTC) |
| Code | Code mode | daily, 00:00 UTC |
| Images | image generation | monthly, on the 1st |
The roleplay pool is the one that most often surprises people: it is weekly, not daily. Spending it all on Monday means no roleplay until the following Monday.
On some plans the monthly image allowance is split — the first portion uses the higher-quality image model, and the remainder falls back to a cheaper one. This is a cost control, and the split for your plan is shown in the product.
2.4 Where to find your numbers
The exact allowance for each pool on each plan is shown in Settings → Usage and on the billing screen, alongside what you have used so far in the current period. We deliberately do not restate those figures here: product limits are tuned more often than a legal document should change, and a stale number in this document would be worse than a pointer to the live one.
2.5 Running out
When a pool is exhausted, that capability stops until the pool resets. There is no overage billing, no borrowing against the next period, and nothing to buy to top it up. You will never receive an unexpected bill from ASAI — the plan price is the whole price.
Unused allowance does not roll over.
3. Billing
3.1 Payment and renewal
Billing is handled by Stripe. Subscriptions renew automatically at the interval you chose until you cancel, which you can do at any time in Settings → Billing.
3.2 Currency and tax
Prices are in EUR. VAT is applied according to Republic of Estonia (EU) rules and the rules of your country. The price shown at checkout includes all applicable tax.
3.3 Cancellation
Cancelling stops the next renewal. Your subscription runs to the end of the period you have already paid for, after which the account drops to Free. Your data is preserved and you can resubscribe at any time.
3.4 Refunds and withdrawal
Your right of withdrawal (14 days) and our refund practice are set out in one place: Refunds & Cancellation. We do not restate the conditions here, so that there is only ever one version of them.
Billing questions: [email protected]. We aim to reply within 5 business days.
3.5 Price changes
We give at least 30 days' notice before a price increase affects an existing subscriber. You can cancel before it takes effect.
4. API access
ASAI has a public HTTP API, and using it is authorised.
- It is OpenAI-compatible at the chat-completions endpoint, so most existing OpenAI client libraries work by changing the base URL and the key.
- Keys are created and revoked in Settings → API.
- API usage draws on the same pools described in Section 2 — it is not a separate allowance and it is not billed separately.
- API access requires a paid plan.
An earlier version of this document stated that programmatic access was not authorised. That was wrong; the API existed when it was written.
We may add separate API terms, pricing, or rate limits later. If we do, they will be published before they take effect.
5. Fair use
The allowances assume normal interactive use. We may act against accounts that:
- Resell access — sharing or selling your account or API key to third parties
- Share credentials — one account, one person
- Deliberately stress the service — consuming capacity without a genuine use
- Attempt to bypass metering — evading the pools by any means
Action scales with severity: throttling, suspension, or termination.
If you have a legitimate high-volume use case, write to [email protected] and we will talk about it rather than throttle you by surprise.
6. Service availability
6.1 No SLA
We aim for high availability and we do not promise a specific uptime percentage on any plan. We are a two-person company; the infrastructure is solid but it is not redundant everywhere.
If an outage lasts more than 24 hours and affects everyone, we may extend affected subscriptions as a goodwill measure. That is discretionary, not contractual.
6.2 Maintenance and beta features
Maintenance is done in off-peak hours where possible and announced in-product. Features labelled beta or preview may change or be withdrawn without notice, and are not grounds for a refund.
6.3 Who runs the models
The models are not ours and do not run on our hardware. ASAI sends your requests to third-party inference providers. This matters for two reasons: it determines who processes your data, and it means model availability is partly outside our control.
Which providers we use, and what they may and may not do with the content we send them, is set out in AI Data & Training and Data Processing & Subprocessors.
We may change the underlying models for cost, capacity, or quality reasons. Replacing a primary model with one from a different family is announced in advance.
7. Suspension and termination
7.1 Grounds
We may suspend or terminate an account for breach of the Acceptable Use Policy, the Community Guidelines, or the Terms of Service; for fraud or chargeback abuse; or where a court order or sanctions regime compels us.
7.2 Effect on your subscription
Suspension does not automatically refund the subscription. If we reverse a decision, we will credit or extend the subscription to make up the lost time.
7.3 Effect on your data
During suspension, data is preserved. After permanent termination, personal data is deleted subject to the schedule in Data Retention, and your published characters and worlds are removed from the library.
You can export your data before deletion — see the Privacy Policy.
7.4 If you think we got it wrong
Write to [email protected] and tell us why. A human reads it and can reverse the decision. As a micro enterprise we are exempt under Article 19 of the EU Digital Services Act from operating a formal internal complaint-handling system, so we are not going to promise you a review board or a decision deadline — but the email is read, and decisions do get reversed. Your right to go to a court or a competent authority is unaffected; see DSA Notice & Action.
8. Changes to these terms
Material changes to commercial terms are announced at least 30 days in advance by email and in-product notice. Continued use after the effective date is acceptance. If you disagree, cancel before that date.
Version history
- 2.0 — 31 July 2026 — Rewritten to match the product. Plans corrected to Free / Pro / Etherane with the roleplay add-ons. Replaced the "seconds of compute × 2" metering description with the actual token-based accounting and the four independent pools, including the weekly roleplay reset. Removed pay-as-you-go and Creator Credits, which do not exist. Corrected the claim that we run the models ourselves. Removed the statement that programmatic access is unauthorised — the API exists and is documented in Section 4.
- 1.0 — 16 May 2026 — Initial publication
Contact
- Billing: [email protected]
- General support and API questions: [email protected]
We accept correspondence in English or Estonian.
Operator
APERTURESyndicate OÜ
Republic of Estonia (EU) · Registry code 17384111 · VAT EE102972654
[email protected]