Usage & Fair Use
Last updated: October 8, 2026
These rules explain how to use AugmentationAI responsibly. They supplement our Terms of Service; our Privacy Policy explains data handling. Mandatory legal rights are not reduced by this page.
What is available now Prepaid balances and pay-as-you-go inference are the current offers. Prices and prepaid balances are held in US dollars (USD). Usage is charged at the rate shown for the selected model or service.
Coding packages and their dedicated router limits are coming soon. No coding-package price, allowance, guaranteed throughput or activation date is offered on this page. Per-minute job billing is not an available purchase unless it is explicitly offered with its unit and price before checkout.
Use the API for legitimate work You may use supported features for lawful development, research and other permitted tasks, provided you have the rights and permissions needed for the material you submit and comply with applicable model-provider terms.
Do not use the service for illegal activity, child sexual exploitation, fraud, credential theft, unauthorized access, harmful malware distribution, or attacks on another person's systems. Legitimate defensive security work on systems you own or have permission to test is not the same as an unauthorized attack. Testing AugmentationAI's own systems requires our prior authorization.
Accounts and keys Protect account credentials and API keys. Do not post keys in public repositories, screenshots or logs, or give an untrusted party access to your store. You are responsible for configuring your applications and agents appropriately.
Do not create extra accounts, pool credentials or rotate identities to evade an applicable published limit or suspension. Do not resell raw credentials or imply that AugmentationAI endorses your product. Any seat, team, customer-facing application or resale restriction will be stated in the applicable offer and the Terms; this page does not turn an API plan into a personal-only seat.
Rate limits are not token allowances Concurrency limits control how many requests can run at once. Request-rate limits control how often new requests can start. Context size, output length and spending limits are separate constraints.
Follow the limits shown for the service you use. If a response includes a Retry-After instruction, wait before retrying. Use bounded retries rather than a tight loop. Do not work around controls by opening more accounts or distributing the same traffic across keys.
No numerical concurrency, request-rate or token allowance is promised here for a future coding plan. Those details, reset windows and any model-specific weighting will be published with that plan before it is sold.
Fair use and traffic management Subject to applicable law and the express commitments of the plan you purchased, we may manage usage at our reasonable discretion to protect service reliability, security, capacity and fair access. Measures may include queueing, adjusting concurrency or request rates, applying reasonable usage limits, changing routing priority, or temporarily restricting a model or feature. Factors may include sustained resource consumption, request patterns, upstream availability, suspected credential sharing or misuse, and effects on other customers.
We may act immediately where reasonably necessary to address abuse, a security incident, a legal obligation or a provider restriction. Otherwise, material changes to a paid offer follow the notice and cancellation provisions in our Terms. Traffic management does not authorize undisclosed charges, remove applicable refund rights, or override an expressly purchased allowance. A restriction is not, by itself, an accusation of misconduct; ordinary capacity controls and enforcement suspensions are different actions.
If we ever describe an offer as unlimited, we will explain any full-speed allowance, reduced-speed operation, queue priority, concurrency limits, model restrictions and reset rules prominently before purchase. We will not use "unlimited" to hide an undisclosed hard quota or a separate pay-as-you-go charge.
Billing, cancellation and refunds Prepaid usage and a monthly package are different products. Do not assume a future coding subscription includes every model or that a memory trial includes paid model usage. The offer shown before purchase specifies what is included.
Top-up minimums, fees and model rates are shown at checkout and match what is recorded in your billing history. Purchased prepaid balances do not expire. Refunds of unused prepaid balances may be requested within 30 days of purchase by contacting support@augmentationai.com, subject to the conditions in our Terms and applicable law. This is a support-request process, not a self-service refund feature. Promotional credits, if offered, state any different conditions when they are given. Statutory refund rights continue to apply.
Cancelling a future subscription, requesting a refund, deleting a store and closing an account are different actions. Consult the Terms and Privacy Policy for the applicable process and data-retention consequences.
Enforcement and review We may restrict requests, revoke affected credentials, or suspend access in response to misuse or a material risk, under the Terms. Where practicable and lawful, we will explain the issue and how to address it. Urgent action may be necessary for security, legal or provider-policy reasons.
If you believe a restriction or charge is incorrect, contact support@augmentationai.com with the request ID, approximate time and a description. Do not include passwords, full API keys, or unnecessary sensitive content. We will review the report; this is not a promise that every provider restriction can be reversed.