Mistral unveils new AI model it says rivals best open systems from China

TL;DR

Mistral has launched a new AI model, Mistral Large 4 (ML4), also called le Chonk, a 1-trillion-parameter model.
ML4 is designed to compete with the best open AI models globally and is claimed to be the strongest open-weight model developed outside China.
The model excels in tasks related to cyber, coding, manufacturing, finance, and multimodal capabilities.
ML4 was trained on 4,000 Nvidia Grace Blackwell GPUs over two months in Mistral's European data centers.
The launch comes amid an intensifying race for technological supremacy between Western companies and Chinese developers in the open AI model market.
Open AI models can be modified and self-hosted, unlike closed systems.
ML4 is being launched in preview before a wider release later this month.
Mistral's co-founder and chief scientist, Guillaume Lample, noted that capabilities will improve with scaled training capacity and highlighted cyber defense benefits.
The article mentions Hugging Face's use of a Chinese model (Z.ai’s GLM 5.2) for defense against a cyber breach, and the unveiling of an open model by Nvidia-backed Reflection AI.

Continuer la lecture de l'article original