Skip to content
ThinkAPI
ProductUsage & costsService statusDocs
Sign inGet started

Terms & policies

Terms of Service ↗Usage Policy ↗Supported Regions ↗Service-Specific Terms ↗
ThinkApiAI Inc312 E 25th St, Cheyenne, WY 82001

Effective date · 2026-05-10

Version · 2026-09-09

Service-Specific Terms

Additional terms for model access, metering, streaming, keys, and data processing.

Service provider · ThinkApiAI Inc

01Models and features02Usage metering and settlement03Streaming, retries, and tool actions04API keys, allowances, and rate limits05Files, multimodal inputs, and optional tools06Request records and request bodies07Payments and support

01Models and features

These terms apply when you use the relevant feature. Specific Claude and GPT versions, input types, context lengths, tools, and protocols depend on the models available to your account and their feature descriptions. Listing a model family does not mean every feature is available.

Providers may modify or retire models. We may update access accordingly and offer explanations or alternatives where feasible. Protocol compatibility does not guarantee identical parameters, error formats, behavior, or outputs to a provider’s direct service.

  • Anthropic Usage Policy ↗
  • OpenAI Usage Policies ↗

02Usage metering and settlement

Charges may include input and output tokens, cache reads and writes, and enabled tools, calculated under the billing rules for the model and service group. Aggregation, conversion, and rounding in the interface may differ from individual displayed items; review request and settlement records together when reconciling charges.

Settlement uses confirmed usage and applicable prices. An error response does not always mean no usage occurred. Missing or unusual metering and billing disputes are reviewed against available records rather than solely whether the client received a complete answer.

03Streaming, retries, and tool actions

Streams may be interrupted by client, network, gateway, or provider failures. Closing a connection does not guarantee that model processing already started stops immediately. Confirmed usage already incurred may still be settled, and resubmitting a request may create a new call and charge.

Clients should retain request IDs, respect retry guidance, and set reasonable retry limits and intervals. Your application should control permissions and necessary confirmations for actions such as running code, changing files, or accessing external systems. Check whether an action has already occurred before retrying it.

04API keys, allowances, and rate limits

A key must be active and assigned to a service group you may use. Key allowances, expiry, account status, model permissions, and service capacity may all restrict requests. A higher spending allowance does not itself grant higher concurrency or reserved capacity.

Use separate keys for different applications and environments, set appropriate allowances, and review usage regularly. Disable or delete compromised keys promptly and inspect unexpected calls. Do not include real keys in tickets or public screenshots.

05Files, multimodal inputs, and optional tools

Files, images, and optional tools are available only when expressly supported by the relevant model and endpoint. Their size, format, and usage limits follow the applicable feature descriptions. You must have a lawful basis to upload materials and necessary permissions for others’ likenesses, voices, personal data, and copyrighted content.

Tools may incur charges beyond model tokens and may connect to third-party services. Review their pricing and permissions before enabling them, and verify returned content and the results of external actions.

06Request records and request bodies

ThinkAPI maintains records relevant to account management, usage settlement, service observation, and security investigation. When enabled, detailed request logging may include request and response bodies. Field redaction cannot guarantee automatic identification of every personal detail or business secret in free text.

The actual scope and retention of logs depend on service configuration, lawful processing purposes, and applicable obligations. Upstream data handling is subject to its own applicable terms and configuration. Unless separately confirmed, do not assume ThinkAPI provides zero logging, zero retention, or compliance with particular data-residency requirements.

07Payments and support

Available payment methods and top-up conditions are shown in the current checkout. A successful payment may take time to appear in your balance. If you have been charged without receiving credit, retain your order number and contact support before paying again.

Redemptions, promotional rewards, and separately agreed enterprise allowances may have their own conditions. Do not use technical means to redeem them repeatedly or evade those conditions. Billing corrections, refunds, and account closure follow the Terms of Service; contact support by ticket or email to resolve these matters.

  • Open a support ticket
  • Email support@thinkapi.org

Contact us with questions about these agreements.

support@thinkapi.org ↗
ThinkApiAI Inc
312 E 25th St, Cheyenne, WY 82001
ThinkAPI

Model APIs, request logs and billing.

ProductFeaturesUsage & costsService statusDocsWorkspace
ContactContact usSubmit a ticket
LegalTerms of ServiceUsage PolicySupported RegionsService-Specific Terms
ThinkAPI
© 2026 ThinkApiAI IncAI API Services