Mistral Medium reaches Claude-level performance on Chatbot Arena (twitter.com)

It's the second-best-performing LLM after GPT-4.

I feel like the title massively underplays the accomplishment. This is now the number two preferred LLM response in a blind comparison, only behind GPT-4. For a model created by a small team in Paris to be ahead of Gemini Pro, Claude, GPT-3.5, etc. is beyond impressive.