CompaRAG (https://comparag.vercel.app/, description: https://www.linkedin.com/pulse/au-coeur-du-premier-comparateur-doutils-rag-fran%C3%A7ais-la-sarazin-phd-rhdje/ and documentation: https://github.com/ArthurSrz/CompaRAG/) compares RAG tools blindly from instructions and documents submitted by the user.
RAG, for “retrieval-augmented generation,” makes it possible to improve the quality of an answer by adding additional resources, see Wikipedia.

CompaRAG is a variation of Compar:IA, focused on RAG tools rather than language models, also on the same principle as Design Arena.

The “Tool Arena” presents in parallel the responses of 2 anonymous tools to the same instruction. The user votes for the best answer before the tools are unmasked.
An overall ranking can be consulted from all the votes collected.

Note: the conversations and votes, anonymized, feed into datasets freely published on HuggingFace.

Free, open-source and account-free.

Test from the school programs of cycle 3 :