← Back to home

Terms of service

These terms form the agreement between you and the AGI Plan operating entity. Section 4 covers regional availability of models — please read it in full.

Last updated
2026-08-14
Status
Draft · not yet in force, pending review by qualified counsel

1Scope and acceptance

  1. 1.1

    These terms are the agreement between you and the operating entity ("we", "us") governing your use of agiplan.dev and the associated API services (the "Service").

  2. 1.2

    Registering an account, placing an order or calling the API means you have read and accepted these terms together with the Privacy Policy, Balance and Billing Terms, Refund Policy, Acceptable Use Policy, Uptime Commitment and Service Commitments. Those documents form an integral part of these terms.

  3. 1.3

    If you use the Service on behalf of an organisation, you represent that you are duly authorised to do so.

2Nature of the Service

  1. 2.1

    We are an independent third-party model aggregation gateway. We purchase inference capacity from multiple upstream model providers and offer it through a single interface with a single unit of measurement.

  2. 2.2

    We have no official authorisation, partnership, agency or endorsement relationship with any upstream model provider, and have never claimed otherwise. Third-party names appearing in the Service are used solely to describe technical compatibility or usage equivalence; all trademarks belong to their respective owners.

  3. 2.3

    We reserve the right to adjust our mix of upstream providers, routing policy and capacity allocation, provided this does not reduce the terms of the capability tier you purchased.

3Accounts

  1. 3.1

    Registration requires a valid email address. Email verification codes are currently the only way to sign in — third-party sign-in (WeChat, GitHub and the like) is not built, and there is no such button in the interface. You must keep your registration information true, accurate, complete and current.

  2. 3.2

    You are responsible for all activity under your account, including custody of API keys. Losses from a leaked key are yours to bear, but you should revoke it in the console as soon as you become aware.

  3. 3.3

    An account is for you personally or for internal use within your organisation. Reselling, renting, or sharing accounts or keys with third parties is a breach, and we may suspend the Service without refund.

  4. 3.4

    We may require identity verification where anti-abuse measures call for it. Accounts requesting refunds must complete identity verification.

4Service territory and model availability

  1. 4.1

    The Service is offered to users worldwide.

  2. 4.2

    Some models have not completed filing or registration for generative artificial intelligence services within mainland China. Under applicable law and regulation, such models are not offered to users located in mainland China.

  3. 4.3

    The console labels the regional availability of each model. Users in mainland China may only use models labelled as available there. We enforce regional restrictions based on registration information, billing information and access origin.

  4. 4.4

    You undertake to state your location truthfully and to ensure your use complies with the law of your own jurisdiction. If you conceal your true location, use a proxy or otherwise circumvent regional restrictions to access a model not offered where you are: that conduct is a material breach of these terms; all resulting legal consequences and liability are yours alone and we bear no responsibility; we may suspend or terminate the Service immediately with no refund for consumed usage; and you shall indemnify us for any loss or regulatory action we suffer as a result.

  5. 4.5

    We reserve the right to adjust model availability by region at any time for compliance reasons. Such adjustment is not a service downgrade, but we will give advance notice and let you switch to another capability tier or receive a pro rata refund.

5Subscriptions and quota

  1. 5.1

    The Service is offered as a monthly or annual subscription. The quota for each plan is the plan version in force when you ordered.

  2. 5.2

    Quota is measured in CU (Compute Units), where 1 CU = 1,000 output tokens on the reference model. Input and output coefficients for every model are published at /models.

  3. 5.3

    Quota is constrained by three concurrent windows: a 5-hour rolling window, the calendar week, and the billing month. The billing month is anchored to your first subscription date, not the calendar month.

  4. 5.4

    Plan versions are immutable. Any change to price, quota or multipliers applies only to new periods, and an increase is announced 30 calendar days in advance. Terms for a subscription you have already purchased do not change within its period.

  5. 5.5

    Unused quota does not roll over or convert to credit at the end of a period. You may cancel at any time and receive a pro rata refund for the unused portion.

6Balance, top-ups and pay-as-you-go

  1. 6.1

    Account balance is divided into paid balance (funds you actually paid) and bonus balance (granted at no cost through promotions, compensation and similar). They differ in nature and are governed separately.

  2. 6.2

    Bonus balance is non-refundable, non-withdrawable, cannot be invoiced, and expires, lapsing automatically at the end of its validity period.

  3. 6.3

    Once plan quota is exhausted, if you have actively enabled pay-as-you-go and set a monthly ceiling, the overflow is charged against your balance at the agreed discount to upstream list price. Bonus balance is drawn first (earliest expiry first), then paid balance.

  4. 6.4

    Pay-as-you-go is off by default. We will not charge anything against your balance unless you turn it on.

  5. 6.5

    Paid balance may be returned through the original payment channel; see the Balance and Billing Terms. We do not offer transfers to third-party accounts, transfers between users, or withdrawals.

7Your data

  1. 7.1

    Zero retention by default. The gateway does not write your request or response bodies to logs, databases or object storage. We record only the metadata metering requires: model identifier, token counts, latency, status code and trace ID.

  2. 7.2

    We will not use your content for any model training or fine-tuning, will not sell it or provide it to third parties in any form, and staff have no access to it. This clause is contractually binding; if we breach it you may demand a full refund and terminate this agreement.

  3. 7.3

    We purchase request corpora from no third-party platform or data broker, and accept no arrangement trading data sharing for purchasing discounts.

  4. 7.4

    We currently offer no body logging of any kind, including a diagnostic log enabled temporarily for troubleshooting — that is, there is no switch anywhere that would let us see your request bodies. Should we ever offer one, it will be off by default, enabled only by you, bounded by a stated retention period, and announced in advance under the change clause of these terms.

  5. 7.5

    See the Privacy Policy for detail.

8Acceptable use

  1. 8.1

    You may not use the Service for unlawful activity, to generate unlawful content, to infringe the rights of others, or to circumvent any upstream provider's usage policy in any way.

  2. 8.2

    You may not: resell the Service; run automated load testing beyond normal use; attempt to bypass quota or billing mechanisms; or reverse engineer or interfere with our systems.

  3. 8.3

    We may suspend the Service immediately for breach of this section. In serious cases we may terminate the agreement without refund.

9Service availability

  1. 9.1

    We undertake to provide the Service at the uptime set out in the Uptime Commitment. Where we fall short, quota credits are issued automatically under that document, with no application required from you.

  2. 9.2

    Planned maintenance is announced in advance. For incidents affecting more than 1% of users we publish a post-mortem within 48 hours.

  3. 9.3

    Unavailability of an upstream provider does not relieve us of our uptime obligations — multi-provider redundancy is our responsibility, not your risk.

10Changes and termination

  1. 10.1

    Material changes to these terms are announced 30 days in advance. If you do not accept them you may cancel before they take effect and receive a pro rata refund.

  2. 10.2

    You may cancel at any time, self-service in the console, without giving a reason. We operate no retention flow.

  3. 10.3

    If we decide to discontinue the Service, we will give 60 calendar days' notice, refund all unused subscription quota and paid balance in full, and assist you in migrating.

11Limitation of liability

  1. 11.1

    The Service is provided "as is". We make no warranty as to the accuracy or fitness of model output — output is generated by upstream models and does not represent our views.

  2. 11.2

    To the maximum extent permitted by law, our total liability to you is limited to the fees you actually paid us in the three months preceding the claim.

  3. 11.3

    We are not liable for indirect loss, loss of profit, loss of data or business interruption.

  4. 11.4

    Nothing in this section excludes or limits liability that cannot be excluded at law.

12Governing law and disputes

  1. 12.1

    These terms are governed by the law of (governing law placeholder).

  2. 12.2

    Disputes arising from these terms shall first be addressed through good-faith negotiation; failing that, they shall be submitted to (arbitral institution / court placeholder).

The full legal name, registration number and ICP filing number of the operating entity appear in the footer. Questions about these terms: support@agiplan.dev.