---
title: "How should I handle an embedding model change in a live search index?"  
description: "How should I handle an embedding model change in a live search index?"  
author: "Marta Kowalska"  
published: 2026-10-21  
updated: 2026-10-21  
canonical: https://www.mindstick.com/forum/162675/how-should-i-handle-an-embedding-model-change-in-a-live-search-index  
category: "Vector Search"  
tags: ["OpenAI", "Embeddings", "Vector Search"]  
reading_time: 1 minute  

---

# How should I handle an embedding model change in a live search index?

I’m updating a semantic search feature and may switch to a newer OpenAI embedding model. The existing vectors are already in a production index, so I’m unsure whether to replace them all at once or maintain separate versions while the new one is tested.

How have you handled an **embedding model upgrade** alongside **incremental document updates**? I’m especially interested in avoiding mismatched vectors during the transition and keeping search available while the index changes.


---

Original Source: https://www.mindstick.com/forum/162675/how-should-i-handle-an-embedding-model-change-in-a-live-search-index

Copyright © MindStick Software Pvt. Ltd. This Markdown version is provided for developers, AI systems, and offline reading.