Top Billing

Legal / Service Level and Support Policy

Service Level and Support Policy

What we aim for on availability and response, stated as targets rather than promises.

Version 1.3 · last updated 2 October 2026

This policy describes what we work to. The Terms of Service govern legally, and clause 12 of them is the warranty position. We would rather publish a target we actually meet than a guarantee we cannot back.

Availability target

We target 99.5% monthly availability for the API gateway, measured as the proportion of successful health checks in a calendar month, excluding announced maintenance and failures inside a third-party model provider that we route around or report as an upstream error.

The current plan carries no service credits. If we introduce a paid tier with a contractual service level and a credit regime, it will be published here, and it will not reduce anything in this document.

Support

Support is by email, at support@topbilling.in, Monday to Friday, 10:00–18:00 IST, excluding public holidays in Tamil Nadu. There is no telephone channel. Quote the tb_req_ request id or the order id and you will usually get a specific answer in the first reply.

SeverityWhat it meansFirst response target
1 — Service downThe gateway is not answering, or authentication or billing is broken for everyone.Within 4 hours, including outside working hours
2 — DegradedElevated errors or latency, one model unavailable, usage or invoices not updating.Within 1 working day
3 — Question or requestHow-to questions, limit increases, a billing query, a document request.Within 2 working days
Billing disputeA charge you believe is wrong.Within 2 working days; investigated within 7
Security reportA vulnerability, under the Responsible Disclosure Policy.Within 2 working days
GrievanceA formal complaint under the Grievance Redressal Policy.Acknowledged within 24 hours, resolved within 15 days

Status and incidents

The status page reports the gateway, the application and each model route from live health data rather than from a manual switch. For an incident of severity 1 or 2 we post what we know, update it as we learn more, and publish a short account afterwards of what happened and what we changed.

Maintenance

Routine deployments are rolling and need no downtime. Where maintenance does need an interruption we announce it on the status page and by email to account owners at least 48 hours beforehand, and we schedule it between 00:00 and 04:00 IST.

Changes and deprecation

  • Model retirement: at least 90 days of notice, with a suggested replacement, unless the provider withdraws it sooner — in which case we tell you as soon as we know.
  • Breaking API change: at least 30 days of notice. Additive changes ship without notice; our protocol surfaces follow the vendor formats they implement.
  • Price change: a new pricing version with an effective time, announced before it applies. In-flight requests keep their version.
  • Policy change: 15 days of notice for anything that materially reduces your rights.
Notices go to the email address on the account. Keep it current.

Other documents