Mistral and NVIDIA team up on AI for industrial engineering
Mistral AI and NVIDIA will combine Mistral Large 4 with accelerated computing to speed up engineering simulation and build public benchmarks for engineering AI agents.
By GennnAi··1 min read
Mistral Large 4NvidiaCUDA-XNVIDIA WarpNVIDIA OmniverseNVIDIA Grace BlackwellNemotron CoalitionMistralModèlesFlash
The work uses NVIDIA's CUDA-X, Warp and Omniverse libraries and targets aerospace, automotive and semiconductors. The benchmarks will measure agents' ability to use tools, run multistep workflows and produce correct results, alongside execution time and cost. Mistral Large 4, with about 1 trillion parameters, was trained on 3,800 NVIDIA Grace Blackwell GPUs; its weights are due at the end of October.
Source: Mistral AI, published on October 21, 2026.