Self-Maintaining Docs

Learn which pages use automation to stay current with live platform data, how current that data is, and how you or an AI agent can check a value before acting on it.

The Hyperstack platform changes constantly. GPU availability shifts from minute to minute, and new hardware, images, and models are added as others are retired. Documentation that lists these values goes out of date quickly, and acting on an outdated value can mean choosing something that is no longer available. The pages below use automation to stay current: their data comes from, or is checked every day against, the live platform, so you can rely on the values you act on.

Data	How current it is	Where it appears
GPU stock	Refreshed at least every 30 seconds while the page is open	GPU Stock Information
AI Studio models	Loaded each time you open the page, from catalog data that refreshes every 2 minutes	Models
Flavors, flavor features, and operating system images	Checked automatically against live data every day, and updated by the documentation team when it changes	Flavors, Flavor Features and Restrictions, Large Root Disk Flavors, and Operating System Images
Public IP address ranges	Checked automatically at least once a week, and updated when it changes	Public IP address ranges

Check a value before you act

You can see how current this data is and where it comes from.

For people

• Status under the title: The GPU stock and Models pages, and the pages kept current by automated data sync, show how current their data is directly under the title. Select the label next to it (Live data or Automated data sync) to see how the page stays current and the API its data comes from.
• The live value: To confirm a value at the moment you act on it, call the source API directly.
• Live stock: The GPU stock dashboard shows how long ago its data last updated.
• Public IP address ranges: The list shows the date the ranges last changed.

For AI agents

AI agents receive the same signals as people reading the pages:

• Markdown pages: The Markdown version of each of these pages includes the same status information: how current the data is and where it comes from.
• Moment-to-moment values: For GPU stock, call the GPU stock API directly. For the AI Studio models available right now, call the List base models API, since the live model table only loads in a browser.

How the data stays current

Data updates on one of three schedules, depending on how often it changes.

Updated live

The dashboard on the GPU Stock Information page shows current figures from the GPU stock API and refreshes at least every 30 seconds while the page is open. The status line above the stock tables shows how long ago the data last updated.

The Available Models table on the Models page loads the current AI Studio catalog from the List base models API each time you open the page, and the catalog data refreshes every 2 minutes. New models and availability changes appear in the table without a documentation update.

Checked daily

Automated data sync checks the data on each page below against live data from the Hyperstack and AI Studio APIs every day. When the live data changes, the documentation team updates the page.

Page	Data kept in sync	Source
Flavors	Every flavor's GPU, CPU, RAM, disk, region, and supported features	List Flavors API
Flavor Features and Restrictions	The feature matrix for every flavor	List Flavors API
Large Root Disk Flavors	Every large root disk flavor and its specifications	List Flavors API
Operating System Images	Every operating system image, with its driver, CUDA, kernel, and region details	List Images API

Checked weekly

Public IP address ranges change infrequently. An automated system checks the public IP address ranges list at least once a week and updates it when the ranges change.