# How Krea AI Accelerates Foundation Model Training with WEKA

Krea AI accelerates foundation model training with NeuralMesh, reaching close to 100% GPU utilization on individual workloads and about 40% MFU on large distributed runs.

**Type:** Case Study

[Download PDF](/api/resource-pdf?slug=how-krea-ai-accelerates-foundation-model-training-with-weka)