Large Enough | Mistral AI

Mistral Large 2 is a 123 billion-parameter model built for single-node inference and supports very long contexts (128k tokens).

It handles dozens of natural languages (including French, German, Spanish, Italian, Portuguese, Arabic, Hindi, Russian, Chinese, Japanese, Korean) and 80+ programming languages such as Python, Java, C, C++, JavaScript, and Bash.

The pretrained model scores 84.0% accuracy on MMLU and improves the performance-to-cost tradeoff among open models.

It significantly outperforms the prior Mistral Large and delivers comparable performance to top models like GPT-4o, Claude 3 Opus, and Llama 3 405B.

Distributed under the Mistral Research License for research and non-commercial use; a Mistral Commercial License is required for commercial self-deployment (contact Mistral).