SentinelGateway
  • Pricing
  • Blog
  • Docs
  • Sign in
  • Dashboard
  • Pricing
  • Blog
  • Documentation
  • Sign In
  • Get Started

Trust & Legal Center

Terms of Service

Effective March 1, 2026 Last updated August 2026

On this page

1. Acceptance & Definitions 2. The Service 3. API Usage & Fair Use 4. Bring Your Own Key (BYOK) 5. Billing, Metering & Spend Protection 6. Availability & SLA 7. Acceptable Use & Abuse 8. Warranty Disclaimer 9. Limitation of Liability 10. Indemnification 11. Termination & Suspension 12. Governing Law & Changes 13. Contact

1. Acceptance & Definitions

These Terms of Service ("Terms") form a binding agreement between you or the entity you represent ("Customer", "you") and SentinelGateway ("Sentinel", "we", "us") governing access to the SentinelGateway AI API gateway, LLM router, Command Center dashboard, and related APIs (collectively, the "Service"). By creating an account, generating an API key, or routing traffic through the Service, you accept these Terms.

"Upstream Provider" means a third-party large-language-model provider (currently OpenAI, Anthropic, Google, and Groq) to which the Service routes inference requests. "BYOK" means configurations where you supply your own Upstream Provider API credentials.

2. The Service

The Service is a zero-trust AI gateway that proxies, secures, caches, meters, and observes inference traffic between your applications and Upstream Providers. Features include multi-provider routing, automatic failover, exact and semantic prompt caching, in-flight PII redaction, prompt-injection and secret scanning, per-tenant token quotas, and usage-based billing.

We may add, modify, or deprecate features with reasonable notice where the change materially reduces functionality you pay for.

3. API Usage & Fair Use

  • Token quotas. Each plan includes a monthly token quota (prompt + completion tokens) displayed in your Command Center. Requests that would exceed your quota are rejected with HTTP 429 until the window resets or you upgrade.
  • Rate limits. Per-tenant request rate limits apply to protect platform stability. Sustained bursting beyond your plan's documented limits may be throttled.
  • Fair use. You agree not to use the Service to generate unlawful, harmful, or rights-infringing content; to probe or disrupt other tenants; or to resell raw access to the Service without a written reseller agreement.
  • Key custody. You are responsible for all activity under your API keys. Rotate keys immediately upon suspected compromise using the rotation endpoint or dashboard.

4. Bring Your Own Key (BYOK) Responsibilities

Where the Service is configured with your own Upstream Provider credentials:

  • Your inference usage is billed by the Upstream Provider directly to you under your own agreement with that provider. SentinelGateway is not a party to that agreement and bears no liability for provider-side charges, rate limits, suspensions, or policy enforcement.
  • You represent that your provider credentials are valid, authorized for your use, and compliant with the provider's terms.
  • SentinelGateway's subscription and metering fees for operating the gateway layer remain due regardless of your provider-side arrangements.

Liability boundaryUnder BYOK, upstream API billing, quota disputes, and provider terms enforcement are strictly between you and your Upstream Provider.

5. Billing, Metering & Spend Protection

  • Subscriptions. Paid plans (Pro, Team, Enterprise) are billed in advance via Stripe and renew automatically until cancelled. Plan changes are pro-rated to the day.
  • Metering. Token usage is recorded per request in your tenant's audit log and metered to Stripe for usage-based components of your plan. Your Command Center spend figures are the authoritative record of billable usage.
  • Failover accounting. When automatic failover serves your request from an alternate model, tokens are metered at the rates of the model that actually served the response, and the fallback event is flagged in your audit trail.
  • Cache hits. Responses served from the exact or semantic prompt cache consume no upstream tokens and are billed at zero token cost.
  • Overages & hard stops. Plans enforce quota hard stops (HTTP 429) rather than surprise overage charges, unless you explicitly enable paid overage capacity.
  • Refunds & taxes. Fees are non-refundable except as required by law. Prices exclude applicable taxes, which will be added where required.

6. Availability & Service Level Commitments

We target 99.9% monthly availability for the gateway data plane, measured excluding scheduled maintenance (announced at least 48 hours in advance) and force majeure events. Enterprise plans may contract a formal SLA with service credits; the 99.9% figure on self-serve plans is a target, not a credit-bearing guarantee.

Because the Service brokers Upstream Providers, end-to-end availability also depends on provider health. Automatic failover is designed to route around provider outages, but we do not warrant the availability, latency, or output quality of any Upstream Provider.

7. Acceptable Use & Abuse

You may not use the Service to:

  • Attack, probe, or scan the gateway infrastructure itself, including deliberate prompt-injection campaigns against the control plane or other tenants.
  • Reverse engineer, decompile, or attempt to extract the Service's source code, routing logic, or another tenant's data, except where such restriction is prohibited by law.
  • Circumvent quotas, rate limits, authentication, or billing mechanisms.
  • Violate any Upstream Provider's usage policies as applied to traffic routed to that provider.
  • Transmit malware, or generate content that is unlawful, defamatory, or infringes third-party rights.

8. Warranty Disclaimer

THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE". TO THE MAXIMUM EXTENT PERMITTED BY LAW, WE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. WITHOUT LIMITING THE FOREGOING, WE MAKE NO WARRANTY REGARDING THE ACCURACY, COMPLETENESS, OR SAFETY OF MODEL OUTPUTS (INCLUDING LLM "HALLUCINATIONS"), NOR THE UNINTERRUPTED AVAILABILITY OF ANY UPSTREAM PROVIDER.

9. Limitation of Liability

TO THE MAXIMUM EXTENT PERMITTED BY LAW: (A) NEITHER PARTY IS LIABLE FOR INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES, OR FOR LOST PROFITS, REVENUE, DATA, OR GOODWILL; AND (B) SENTINELGATEWAY'S AGGREGATE LIABILITY ARISING OUT OF THESE TERMS IS LIMITED TO THE FEES YOU PAID US IN THE TWELVE (12) MONTHS PRECEDING THE EVENT GIVING RISE TO THE CLAIM.

We are not liable for outages, errors, rate limiting, content filtering, or policy actions of Upstream Providers; for model output errors or hallucinations; or for losses arising from your failure to safeguard API keys.

10. Indemnification

You will defend, indemnify, and hold harmless SentinelGateway from third-party claims arising out of (a) your applications, content, or prompts routed through the Service; (b) your violation of these Terms or of an Upstream Provider's terms; or (c) your misuse of the Service. We will promptly notify you of any indemnified claim and reasonably cooperate in its defense at your expense.

11. Termination & Account Suspension

  • By you. You may cancel at any time from the Billing portal; cancellation takes effect at the end of the current billing period.
  • By us — for cause. We may suspend or terminate access immediately for abuse (Section 7), non-payment, security risk to the platform or other tenants, or material breach of these Terms.
  • Effect. Upon termination, your API keys are revoked and your quota counters stop. Audit logs and account records are handled per the retention schedule in our Privacy Policy.

12. Governing Law & Changes

These Terms are governed by the laws of the State of Delaware, USA, without regard to conflict-of-law rules. Disputes will be resolved in the state or federal courts located in Delaware, and the parties consent to their exclusive jurisdiction.

We may update these Terms from time to time. Material changes will be announced by email and by a prominent notice in the Command Center at least 14 days before taking effect. Continued use after the effective date constitutes acceptance.

13. Contact

Questions about these Terms: legal@sentinelgateway.ai. General support: support@sentinelgateway.ai.

Privacy Policy · Terms of Service · Contact Support
SentinelGateway
© 2026 SentinelGateway  ·  Terms · Privacy Policy
Product
  • AI Gateway
  • Semantic Cache
  • PII Scrubbing
  • Request Tracing
  • Pricing
Legal
  • Privacy Policy
  • Terms of Service
  • Contact Support