Skip to content Skip to footer

Cohere AI introduces Rerank 3: An innovative base model created to enhance enterprise search and enhance Retrieval Augmented Generation (RAG) systems.

Artificial Intelligence (AI) company Cohere has launched Rerank 3, an advanced foundation model designed to enhance enterprise search and Retrieval Augmented Generation (RAG) systems, promising superior efficiency, accuracy, and cost-effectiveness than its earlier versions.

The key beneficiaries of Rerank 3 are enterprises grappling with vast and diverse semi-structured data, such as emails, invoices, JSON documents, code, and tables. The software’s improved management of various metadata fields, including data recency, enables pertinent search results. Besides, unlike its predecessors, Rerank 3 supports more than 100 languages, catering to the global organizations’ multilingual data sources. This feature not just simplifies data recovery across languages but also encourages inclusivity and utility for non-English speaking users.

Integration of Rerank 3 with existing search systems or legacy applications is simple, requiring just a single line of code. Furthermore, the model’s ability to process longer documents, up to 4k context length, removes the need for data chunking, ensuring the preservation of the full context of search queries. The model’s enhanced capabilities also extend to processing data types such as JSON and tabular data in databases, CSVs, and Excel sheets, which are fundamental for enterprises but traditionally difficult for retrieval models.

Rerank 3 delivers three times lower latency in search operations in contrast to its earlier version, making it apt for time-sensitive applications in sectors like e-commerce and customer service. Integration with Cohere’s Command R series models for RAG applications results in substantial total cost of ownership (TCO) reduction, up to 98% less than other large language models (LLMs), without reducing quality.

Notable features of Rerank 3 include improved efficiency through complex data handling and multilingual retrieval, a substantial reduction in costs due to integration with Cohere’s Command R models, better accuracy, and lower latency, ease of integration, and broader scope of searchable enterprise data due to its ability to process semi-structured data such as JSON and tabular data.

Rerank 3 is available on Cohere’s hosted API and AWS Sagemaker. It’s also accessible through the inference API in Elasticsearch to conduct semantic reranking on your existing Elasticsearch index. With its considerable enhancements and features, Rerank 3 necessarily becomes an essential tool for any data-driven organization seeking to optimize its information retrieval processes.

Leave a comment

0.0/5