Mistral Large 2 has improved performance in terms of cost and serving on evaluation metrics, with an accuracy of 84.0% on MMLU. It has enhanced reasoning capabilities and is trained to acknowledge when it cannot find solutions or lacks sufficient information. It also has improved instruction-following and conversational capabilities and is better at following precise instructions and handling long multi-turn conversations. The model has been trained on a large proportion of multilingual data, making it suitable for business use cases involving multilingual documents. It is available on la Plateforme under the name `mistral-large-2407` and can be accessed through leading cloud service providers.
Key takeaways:
- Mistral Large 2 is a new generation AI model designed for single-node inference with long-context applications, supporting dozens of languages and 80+ coding languages.
- The model has been trained to minimize the tendency to generate incorrect or irrelevant information, and is designed to acknowledge when it cannot find solutions or does not have sufficient information to provide a confident answer.
- Mistral Large 2 has enhanced function calling and retrieval skills, and has undergone training to proficiently execute both parallel and sequential function calls, making it suitable for complex business applications.
- Mistral Large 2 is available on la Plateforme under the name `mistral-large-2407`, and can be accessed through leading cloud service providers including Google Cloud Platform, Azure AI Studio, Amazon Bedrock and IBM watsonx.ai.