Description
Ragas is an open-source evaluation framework for AI applications, especially those built on retrieval augmented generation (RAG) and AI agents. It gives you metrics to score how well your application performs, including context precision and recall, faithfulness, response relevancy, noise sensitivity and answer accuracy, and it also has metrics for agent and tool use. You prepare datasets, run experiments and check the results against those metrics. The docs include quick starts and tutorials for evaluating a prompt, a simple RAG system, an AI workflow and an AI agent. The project is on GitHub under vibrantlabsai/ragas. The site also runs office hours for teams that want help setting up evals for their AI application. The docs are organised into Get Started, Core Concepts, How-to Guides, References and Community.…
Highlights
- Metrics for RAG and agent applications
- Faithfulness, precision and recall scores
- Datasets and experiments
- Tutorials for prompts, RAG and agents
- Open source on GitHub
Ratings & Reviews
No ratings yet
Be the first to rate RAGAS.
People also visit
RAGAS FAQ
What is Ragas?
An evaluation framework for AI applications. It scores how well a RAG system or an AI agent performs, using a set of metrics.
Which metrics does it have?
Retrieval metrics such as context precision, context recall and noise sensitivity, generation metrics such as faithfulness and response relevancy, and metrics for agent and tool use.
How do I get started?
The docs have Installation and Quick Start pages. Start there, then follow the tutorials for your kind of application.
Is it free?
The project is open source, and its code is on GitHub at vibrantlabsai/ragas.
Is there help with setup?
The site runs Office Hours for teams that want help setting up evals for their AI application.
What are the best RAGAS alternatives?
Popular alternatives to RAGAS include Ragbuilder, Krux AI's RAGBuilder and ragie.ai. See all RAGAS alternatives












