Inspekt
Get a key

Service Level Agreement

This is a superseded version. It is the Service Level Agreement as published on 17 September 2026, kept here unchanged because the change history says superseded text stays readable. The version in force is the current Service Level Agreement.

We commit to 99.5% monthly uptime. Section 3 states what is behind that number.

1. Who this covers

1.1 This SLA applies to the embed, Render API and Inventory API on any plan with a monthly subscription fee: Hobby, Pro and Scale, unless an account's written agreement says otherwise.

1.2 Enterprise, which is pay as you go, is best effort, with no uptime commitment and no credits, unless its written agreement says otherwise. A credit is a percentage of the subscription fee for the affected month and Enterprise has no subscription fee. An account with no plan (Terms 5.7) is not served at all and is outside this SLA for the same reason.

2. The commitment

2.1 We commit to 99.5% monthly uptime, calculated as:

Monthly Uptime % =
    (total minutes in the calendar month - downtime minutes)
    / total minutes in the calendar month
    x 100

measured per calendar month, UTC. The allowance is 0.5% of the month, so it depends on the month: about 3 hours 36 minutes across 30 days, and 3 hours 43 minutes across 31.

2.2 Downtime means minutes in which most well-formed requests to the embed, Render API or Inventory API return a server error (HTTP 5xx) or fail to answer within 30 seconds, measured under section 4, excluding anything covered by section 5.

3. What the service actually runs on

3.1 Inspekt runs from a single rented server in a United States datacentre, with no automatic failover. There is no redundant hosting behind the service. A hardware failure, network fault or provider outage at that server takes the whole thing offline. Static assets reach you through a content delivery network (CDN) in front of that server. New images from the Render API are drawn on one self-hosted GPU machine on a residential internet connection. While that machine is unreachable, every image already rendered is still served, and a request for one that has not been rendered yet gets a placeholder image until the machine is back.

3.2 We say this outright so you can decide whether it suits what you are building. The 99.5% figure in section 2.1 was set with this infrastructure in mind.

3.3 We may improve this over time, for example by adding a second server or a second GPU machine. Any such change can only improve on what section 3.1 describes, and nothing here obliges us to make one.

4. How we measure it

4.1 Our own monitoring is the authoritative source for downtime minutes. We run synthetic checks against the public surfaces every 5 minutes. We have no independent third-party monitoring; see section 4.2.

4.2 If you make a credit claim under section 8, we will give you the monitoring data we hold for that month on reasonable request.

5. What does not count as downtime

5.1 Scheduled maintenance, announced at least 48 hours ahead by email to your account contact, capped at 4 hours in a calendar month, and performed outside 09:00 to 21:00 US Mountain Time where that is reasonably practicable.

5.2 Your own or an end user's equipment, software, network or connectivity, or any third-party service you depend on other than ours.

5.3 Your breach of the agreement, misuse of your key, or traffic well beyond your plan or any reasonable expectation of it.

5.4 Force majeure: acts of God, war, terrorism, civil unrest, government action, internet backbone or DNS root failures, a widespread cloud or ISP outage outside our control, pandemics and similar events.

5.5 Valve changing, restricting or discontinuing access to the CS2 game files, item schema or Steam services we depend on, or any other third-party dependency going away or changing in a way that breaks us.

5.6 Features we have explicitly labelled beta, preview or experimental.

5.7 A denial-of-service or similar attack we mitigate in good faith, even where the mitigation itself causes some unavailability.

6. Service credits

If monthly uptime falls below 99.5%, you are entitled to a credit against that month's subscription fee, not against overage:

Monthly uptimeCredit
99.0% to under 99.5%10% of that month's subscription fee
95.0% to under 99.0%25%
90.0% to under 95.0%50%
Under 90.0%100%

Credits for a single month are capped at 100% of that month's subscription fee.

7. New content

The commitment to render newly released CS2 content within 7 days is separate from uptime and lives in Terms section 9.

8. Claiming a credit

8.1 Email hello@inspekt.gg within 30 days of the end of the affected month, with the dates and approximate times you saw downtime.

8.2 We check the claim against our monitoring under section 4 and, if it holds up, apply the credit to a future invoice. Credits are not paid in cash, are not transferable, and expire 12 months after they are issued.

9. Sole remedy

Service credits are your only remedy for us missing the uptime commitment in section 2.1. That limits remedies for downtime specifically. It does not limit your other remedies for a breach of the agreement generally, which remain subject to Terms section 13.

10. Changes

We update this SLA the same way we update the Terms, under Terms section 16.