The LMSYS Chatbot Arena service has updated the rating of large language models of artificial intelligence. It allows you to evaluate which models are currently doing the best job.
LMSYS Chatbot Arena is a crowdsourced open platform for evaluating large language models (LLMs). To compile the rating, more than 300 thousand human reviews of the models’ performance are evaluated using the Elo rating system.
How the test works — people enter a query and choose the best answer from several options from different models. Thousands of user tests are used to formulate and rank the top models.
According to the new rating of the chatbot arena, GPT4 is currently the leader among LLMs. Claude’s recent claims that their model is better have not been confirmed. It was ranked third. Right behind it is the Bard (Gemini Pro) model from Google. All these models received an Elo rating of over 1200.
You can find detailed up-to-date ranking results of the existing major language models at to the following address.