Google Cloud
Vertex AI retires text-embedding-004/005 and legacy embedding models on April 1, 2027
133 days left — Apr 01, 2027
- What stops working
- On April 1, 2027 Vertex AI retires text-embedding-005, text-embedding-004, text-multilingual-embedding-002 and multimodalembedding@001. Calls to these model IDs stop working. Vector indexes built with them cannot be queried with embeddings from a different model, so every corpus has to be re-embedded.
- Who is affected
- Any retrieval, search or recommendation system on Google Cloud that still creates embeddings with these model IDs.
- Official migration path
- Move to a current Gemini embedding model and re-embed existing content before the retirement date.
- What the official path does not cover
- Google does not re-embed or migrate existing vector stores; switching models changes the vector space, so all indexes must be rebuilt, re-evaluated for retrieval quality and cut over without downtime.
- Source
- Vertex AI: Model versions and lifecycle (last updated Nov 18, 2026) · announced 2026-11 · verified 2026-11-19
Need this done before Apr 01, 2027?
Fixed price, fixed scope, fixed date. From $4,000 for a single integration or workflow; you get the working replacement, the code, and a before/after check. Quote within one business day.
One short form (company, email, what runs on this, needed-by). Reply within one business day.
Not ready yet? Get an email when Google Cloud deadlines approach · Firms: get listed for Google Cloud work.
Spotted a wrong date? Report it — corrected with the source within one business day.