Elastic announced on 18 November that its jina-embeddings-v5-text and jina-embeddings-v5-omni models have joined the Microsoft Foundry model catalog

1

2

. The models deploy as dedicated endpoints inside customers' own Microsoft Azure environments, so developers can run them in their own Azure accounts rather than through a shared external service

1

2

.

What shipped

The Jina embeddings v5 family spans four models in two groups. The text pair, jina-embeddings-v5-text-small and jina-embeddings-v5-text-nano, produces multilingual text embeddings. The company puts their language coverage at more than 100, spanning retrieval, matching, clustering and classification work

1

2

. The omni pair, jina-embeddings-v5-omni-small and -nano, adds images, audio and video in a single unified vector space

2

.

Elastic's stated aim is that teams building AI agents and search apps on Azure can add multilingual, multimodal retrieval without moving their data

1

. "AI applications succeed or fail on retrieval, and the best embedding model is the one a team can run where its data already lives," says Han Xiao, Elastic's vice president of AI

1

.

The vendor's case

Jina AI, which Elastic owns, builds small, purpose-built models that the company says sit at or near the top of the benchmarks it cites

2

. Per Elastic, the omni models match or beat models up to 20 times their size on image benchmarks

2

. The company puts them ahead of their size class on audio and says they hold their own against far larger models on video grounding

2

.

The vendor also says the models bring semantic search to Azure applications without the compute costs that larger models carry

2

.

© 2026 TheOutpost.AI All rights reserved