> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getcited.in/llms.txt
> Use this file to discover all available pages before exploring further.

# Model Deprecation

> When an AI provider retires or replaces a model version, potentially changing brand visibility behavior.

**Model deprecation** is when an AI provider retires or replaces a specific model version — for example, when OpenAI moves from one ChatGPT generation to the next, or when Perplexity upgrades from one Sonar version to another. Model deprecations can shift brand visibility behaviour without any change in brand content, because the new model may have different training data, different response patterns, and different source preferences.

## Why it matters

A model update can cause a brand's [mention rate](/glossary/mention-rate) to change overnight — not because the brand did anything differently, but because the underlying AI model changed how it generates answers. This is one of the [noise sources](/methodology/data-freshness) that Cited monitors. Brands that see sudden metric shifts should check whether the AI provider recently updated its model before investigating content-level causes.

## How Cited uses it

Cited tracks which platform version is used for each AI engine and documents changes when they occur. When a model is deprecated, Cited validates the replacement against historical data before switching, so your dashboard trends remain comparable. The currently tracked platforms are documented in [Which AI platforms we track](/methodology/which-llms-we-track). Model changes are one of the known sources of measurement [non-determinism](/glossary/non-determinism).

## Related concepts

* [Which AI platforms we track](/methodology/which-llms-we-track) — the tracked platforms and the rationale behind the set
* [Non-determinism](/glossary/non-determinism) — model updates as a source of response variance
* [Reading your data with confidence](/methodology/data-freshness) — interpreting metric shifts in context