FrontierThe story, in brief
Microsoft AI Releases Harrier-OSS-v1: A New Family of Multilingual Embedding Models Hitting SOTA on Multilingual MTEB v2
Microsoft just open-sourced three embedding models that hit SOTA on multilingual benchmarks. Here's why semantic search just got cheaper.
The KeyNews take
Why it matters
Microsoft's Harrier-OSS-v1 family (270M to 27B parameters) achieves state-of-the-art multilingual performance, signaling a competitive push in the embedding model space where cost and cross-language capability are becoming table stakes for enterprise AI.
The key facts
5 to knowThree model scales: 270M, 0.6B, 27B parameters
Achieved SOTA on Multilingual MTEB v2 benchmark
Open-source release (OSS designation)
Multilingual semantic representation capability
Published March 30, 2026
Go to the source
MarkTechPostmarktechpost.com
Publisher excerpt: Microsoft has announced the release of Harrier-OSS-v1, a family of three multilingual text embedding models designed to provide high-quality semantic representations across a wide range of languages. The release includes three distinct scales: a 270M parameter model, a 0.6B model, and a 27B model.…