Mistral Large 4 enters Agent Arena's top 15 labs at -6.6% net improvement score

Arena (@arena) reports that Mistral Large 4, a preview model from Mistral AI, ranks #43 overall in its Agent Arena with a -6.6% net improvement score across more than 5,000 real-world agentic sessions.

Arena says this places Mistral among the top 15 labs, the only European lab there, and 11 places above its predecessor, Mistral Medium 3.5, which scored -12.60%.

Arena also says that Mistral Large 4's open weights are expected at the end of October, and that at its current score the model would rank #13 among open models. These are Arena's own figures and projections; the open-weights ranking is conditional on the release and on the score holding.