Granite R2 Delivers Top Multilingual Embeddings

IBM released the Granite Embedding Multilingual R2, featuring a compact 97M-parameter model and a full 311M-parameter model with superior multilingual retrieval capabilities.

This release enhances multilingual and code retrieval capabilities without significantly increasing model size, contributing to more efficient and accessible AI tools.

The new models benefit AI developers and multilingual teams requiring robust retrieval in multiple languages and codebases.

It remains uncertain how these models will perform in diverse real-world applications where language nuances and large data demands are factors.