title: "Un Ministral, des Ministraux | Mistral AI"

Mistral AI has introduced two new state-of-the-art models, Ministral 3B and Ministral 8B, designed for on-device and edge computing use cases. These models are optimized for compute-efficient, low-latency solutions in applications like local translation, autonomous robotics, and agentic workflows. Both models support a large context length, with the 8B version featuring a special attention pattern for memory-efficient inference. Benchmark results demonstrate that these new models consistently outperform their peers in the sub-10B category, including models from the Llama and Gemma families.