AIMistral Blog58m ago
Mistral says ML4 was trained using 3,800 Nvidia Grace Blackwell GPUs
Mistral says ML4 was trained using 3,800 Nvidia Grace Blackwell GPUs in its own data centers in Europe and much of its training data was multilingual
TL;DRMistral trained a new large language model using thousands of GPUs in European data centers with multilingual data.
Why it matters: Shows EU AI independence from US cloud providers; multilingual focus could challenge US-dominated models in global markets.
Le Chonk — Today, we're launching a public preview of Mistral Large 4. Unofficially ML4, very officially: le Chonk.
Read full articleSource: Mistral Blog · Opens in new tab