>
Today, we are announcing Mistral Large 2, the new generation of our flagship model. Compared to its predecessor, Mistral Large 2 is significantly more capable in code generation, mathematics, and reasoning. It also provides a much stronger multilingual support, and advanced function calling capabilities.
Notes: Details are sparse, but we can hazard a guess based on evidence about the training cluster they may have used, the scale up in compute they likely would have used relative to Mistral Large 1, and from the model's MMLU score. Extended reasoning given here: https://docs.google.com/document/d/1I2ZWBLFMpRZYcdMMUfKAGZFJrOJpduNDS9ZeVFIHnd8/edit?usp=sharing