AB
AiBoss
project

Mistral Small 3 - An open-source large language model from Mistral AI

Mistral Small 3 is an open-source large-scale language model released by Mistral AI, containing 24 billion parameters and licensed under the Apache 2.0 license. It is designed for low-latency tasks and supports multiple languages (including English, Chinese, Japanese, etc.).

What is Mistral Small 3?

Mistral Small 3 is an open-source large-scale language model released by Mistral AI, containing 24 billion parameters and licensed under the Apache 2.0 license. Designed for low-latency tasks, it supports multiple languages (including English, Chinese, Japanese, etc.) and excels in inference and multilingual processing. The model employs a Transformer architecture with 40 layers, a 32k context window, and supports long inputs. Optimized for local inference, it can run on laptops with an RTX 4090 GPU or 32GB of RAM, making it suitable for conversational AI, automation tasks, and specialized domain applications.

Main functions of Mistral Small 3

  • Low latency optimizationThe Mistral Small 3 is designed for low-latency tasks and is suitable for scenarios requiring rapid response, such as virtual assistants, automated workflows, and real-time AI applications. It offers inference speeds more than 3 times faster than the Llama 3.3 70B on the same hardware.
  • high performanceWith 24 billion parameters, its performance rivals that of larger-scale models such as Llama 3.3 70B and Qwen 32B. In MMLU benchmarks, it achieves over 81% accuracy with a latency of only 150 tokens/s.
  • Multilingual supportIt supports multiple languages, including English, French, German, Spanish, Italian, Chinese, Japanese, and Korean.
  • Local deployment capabilitySuitable for scenarios with high data privacy requirements, it can run on a single NVIDIA RTX 4090 graphics card or a MacBook with 32GB of memory.
  • Open source and customizabilityIt is open source under the Apache 2.0 license, allowing developers to freely modify and deploy it, making it suitable for enterprise applications.
  • Command following and reasoning abilityWith fine-tuning of instructions, it can understand and execute various tasks, including code generation, mathematical calculations, and general knowledge questions and answers.

Mistral Small 3 project address

Application scenarios of Mistral Small 3

  • Virtual AssistantMistral Small 3 can act as a virtual assistant, quickly responding to user needs and providing natural language interaction.
  • Customer service systemIt can be integrated into the customer service system to automatically answer common questions and improve the user experience.
  • Workflow Automation: Execute tasks quickly in automated workflows to improve efficiency.
  • Medical diagnosisWith fine-tuning, it can be used in medical diagnostic systems to quickly generate diagnostic suggestions.
  • Legal ConsultationProvides professional advice in the legal field to help users answer legal questions.