Terms of Service
Effective and last updated: October 19, 2026
These policies describe how GhostRoute provides access to third-party AI models, handles prepaid usage and processes information. Questions and requests are handled privately inside your workspace.
Terms of Service
1. The service
GhostRoute is a technical interface to third-party models. Model availability, context limits, supported features and current prices appear in the catalog. Outputs may be inaccurate, incomplete or unsuitable for a particular purpose. Review outputs before relying on them. Reduced refusal behavior does not guarantee a particular response.
2. Workspace access
You can sign in with a username and password or use a key-only workspace. Your workspace recovery key is a separate way to access your balance, API keys and request history. Keep your password and recovery key private, and store a separate recovery copy. API keys authorize generation subject to their limits. You are responsible for requests authorized by your keys; revoke a compromised key promptly. Lost access may require additional ownership verification before records or funds can be released.
3. Prepaid usage and prices
Balances are denominated in USD and can be funded with USDT (TRC20) or TRX on TRON. The invoice specifies the required asset, network and exact amount. TRX is converted to a USD credit using the exchange rate captured when the invoice is created. TRX quotes are valid for 20 minutes; later payments require manual review. Credit requires full payment and 20 network confirmations. Network charges are separate. A workspace balance is prepaid service credit, not a bank account, investment or interest-bearing deposit.
New username accounts receive a one-time $1 promotional API credit. This credit can only pay for service usage; it is not a customer payment and cannot be withdrawn, transferred or refunded. Signing in, recovering access or changing credentials does not award another bonus. Creating multiple accounts to repeatedly claim welcome credit is not permitted.
Input and output prices are per one million tokens. We reserve an upper cost estimate before accepting a generation, settle against confirmed usage and release the remainder. Each charge rounds up to the nearest millionth of a dollar. Reasoning and cached tokens follow the published input/output rates; no separate cache discount is currently promised. Prices captured when a request is accepted govern that request; later price changes apply to new requests.
4. Failed and uncertain requests
An explicit provider rejection releases the hold without a generation charge. If the connection or usage record is incomplete, the request may remain on hold pending reconciliation. A transport error does not prove that the provider performed no work. Do not automatically repeat an uncertain request. A repeated Idempotency-Key is rejected without starting another generation.
5. Responsible use
Use the service lawfully and respect the rights of others. Do not use it to exploit children, conduct fraud, distribute malicious software, unlawfully access systems, violate privacy or evade applicable restrictions. Do not probe other workspaces or attempt to bypass billing or security controls. Relevant model-provider terms also apply. Access may be restricted to investigate abuse, protect the service or comply with legal requirements.
6. Availability and changes
Third-party models and networks may become unavailable. The service does not promise uninterrupted access or a particular model indefinitely. Changes to these policies will be published with an updated effective date. Changes do not retroactively remove an accrued refund entitlement or a non-waivable right.
7. Questions and mandatory rights
Use Requests in your workspace for service questions, billing disputes or privacy matters. These terms do not exclude remedies or consumer rights that applicable law does not allow us to exclude.
Open a private request ↗