Un Ministral, des Ministraux

Mistral AI · 1 year ago

Mistral AI released two new small language models, Ministral 3B and Ministral 8B, designed for on-device and edge computing tasks like local translation, smart assistants, and robotics. Mistral 3B costs $0.04 per million tokens and Ministral 8B costs $0.1 per million tokens on their platform, with both supporting 128k context length. The models are positioned as efficient intermediaries for multi-step workflows and outperform larger competitors on standard benchmarks according to Mistral's internal evaluations.