Chrome's Text Embedding Model Shrinks by 57%, Maintains Performance
2025-05-13
Chrome's latest update includes a new text embedding model that's 57% smaller (35.14MB vs 81.91MB) than its predecessor, while maintaining virtually identical performance in semantic search tasks. The size reduction was achieved through quantization of the embedding matrix from float32 to int8 precision. This optimization significantly improves storage efficiency for Chrome, particularly beneficial on devices with limited storage, such as budget smartphones and tablets, without sacrificing search quality.
Read more
(dejan.ai)
Tech
Semantic Search