# OpenAI Announces New Embedding Model Trained with Contrastive Learning

OpenAI has revealed a new approach to creating embeddings for text and code using contrastive pre-training methods.

The announcement, shared via the company's official Twitter account, signals an advancement in how AI models understand and represent both natural language and programming code in a unified way. Embeddings are mathematical representations that capture the meaning of text or code, allowing AI systems to compare, search, and reason about information.

Contrastive pre-training is a machine learning technique that teaches models to distinguish between similar and dissimilar examples. By applying this method to both text and code simultaneously, OpenAI's new model can better understand the relationships between human language and programming languages.

This matters for developers and businesses because better embeddings lead to more accurate semantic search, improved code completion tools, and more effective retrieval systems. The dual focus on text and code is particularly