Vectara has launched Open RAG Eval, an open-source framework designed to evaluate Retrieval-Augmented Generation (RAG) systems. This framework aims to systematically assess various components of RAG, including retrieval, generation, citation, and hallucination. The initiative is a response to the challenges developers face in evaluating RAG applications, particularly as most existing frameworks are tailored for large language models (LLMs) rather than RAG systems. Vectara collaborated with Professor Jimmy Lin and the University of Waterloo to develop this comprehensive evaluation tool. The launch comes amid discussions within the industry regarding the effectiveness of current RAG applications, with some reports indicating that enterprise-grade RAG systems have struggled with schema-aligned queries, achieving a score of 0% in accuracy benchmarks. Other companies, such as Qdrant and HoneyHive, are also working on tools to enhance the evaluation and performance of RAG systems, focusing on diagnostics and optimization.