As reported by TechCrunch, the French AI firm Mistral has recently rolled out its Mistral 3 series of models. This lineup encompasses 10 open-source weight models: the flagship multimodal large-scale model, Mistral Large 3 (boasting 41 billion active parameters and a staggering total of 675 billion parameters), alongside nine more compact Ministral 3 models (ranging from 3 billion to 14 billion parameters). All these models are designed to support multilingual functionalities, visual input, and extended-context processing. They are distributed under the Apache 2.0 open-source license, with their weights made available on both Hugging Face and GitHub platforms. Mistral underscores that, following fine-tuning, the smaller models present significant cost and deployment benefits for enterprises, being capable of operating on single GPUs or even edge devices. Presently, Mistral is forging partnerships with entities like HTX, Helsing, and Stellantis to drive the practical deployment of AI in areas such as robotics, security systems, and in-vehicle intelligence.