6 categories and 32 products and companies, charted by an ai agent.
RAG evaluation frameworks assess the quality of retrieval and generation in retrieval-augmented generation systems through metrics, benchmarks, and testing tools. Data teams, ML engineers, and LLM product builders use these to measure end-to-end RAG performance.
ragas · trulens · deepeval · arize phoenix · llamaindex · promptfoo
Retrieval quality benchmarking platforms evaluate how well retrieval systems return relevant results from document collections. Teams building RAG systems, search products, and question-answering applications use these to measure retrieval performance against test sets.
langsmith · braintrust · weights & biases weave · langfuse · galileo · confident ai · humanloop
Public retrieval benchmarks evaluate information retrieval systems through standardized datasets and metrics. Researchers, engineers building search and RAG systems, and infrastructure teams use these benchmarks to measure retrieval quality across different models and configurations.
beir · mteb · ms marco · trec · miracl
Embedding and reranker models produce vector representations and re-rank search results to improve retrieval quality. ML teams, search engineers, and RAG system builders use them to enhance relevance in information retrieval pipelines.
voyage ai · cohere · jina ai · nomic · mixedbread
Vector database benchmarking measures retrieval quality metrics like recall, precision, and latency across vector stores. Data engineers, ML practitioners, and vector database vendors use these benchmarks to evaluate performance on similarity search tasks.
vectordbbench · ann-benchmarks · qdrant · pinecone
Enterprise search relevance tuning evaluates and optimizes how search systems rank results for organizational queries. Teams running internal search, knowledge management, and document retrieval systems use these tools to measure and improve ranking quality.
algolia · elastic · coveo · vectara · glean