收购 Emmi AI 凸显了 Mistral 致力于推动人工智能研究与工业工程企业解决方案的前沿发展。
Emmi 的工作现已并入 Mistral，其核心目标是从根本上帮助工程师更快地构建下一代产品，并为客户在规模化运营中持续获得性能提升提供保障。
我们正加倍投入，为那些塑造物理世界的行业（如航空航天、汽车、半导体和能源）构建基础物理 AI。
以下是这项研究工作所依托的部分已发表突破性成果。
2025 年 12 月 1 日
与音速同行：将神经代理模型推向高湍流跨音速领域
现有的航空航天数据集主要关注二维翼型，忽略了这些关键的三维现象。为弥补这一空白，我们提出了一个全新的跨音速三维机翼 CFD 模拟数据集。该数据集包含约 30,000 个样本的体积场和表面场数据，每个样本具有独特的几何形状和入流条件。
arXiv
2025 年 11 月 25 日
流体智能：计算流体力学中 AI 基础模型的前瞻
在 GPU 和 AI 进步的推动下，计算流体力学领域正在经历重大变革。本文通过将工业级 CFD 模拟分解为其核心组成部分，弥合了机器学习与 CFD 社区之间的鸿沟。
arXiv
2025 年 10 月 17 日
面向汽车与航空航天应用的 AB-UPT
在本技术报告中，我们为 AB-UPT 的经验评估用例库新增了两个数据集，将高质量数据生成与最先进的神经代理模型相结合。
arXiv | Github
2025 年 10 月 8 日
GyroSwin：用于回旋动理学等离子体湍流模拟的五维代理模型
核聚变在寻求可靠且可持续的能源生产中扮演着关键角色。实现可行聚变能的一个主要障碍是理解等离子体湍流，这种湍流会严重削弱等离子体约束，对于下一代反应堆设计至关重要。
arXiv | Github
2025 年 2 月 23 日
AB-UPT
锚定分支通用物理 Transformer（AB-UPT）用于空气动力学 CFD。可在单张 GPU 上处理 900 万表面单元和 1.4 亿体积单元的原始几何体，无需重新网格化。
arXiv | Github
2024 年 11 月 14 日
NeuralDEM
首个面向大规模多物理过程的端到端深度学习代理模型。能够对工业过程（如流化床反应器）进行实时仿真。
arXiv | Github
2024 年 2 月 19 日
UPT：通用物理 Transformer
一种高效扩展神经算子以处理多种时空问题的框架。支持网格和粒子两种仿真方式。
arXiv | Github
The acquisition of Emmi AI has highlighted Mistral’s commitment towards pushing the state-of-the-art in AI research and enterprise solutions for industrial engineering.
Emmi’s work, now part of Mistral, is dedicated to fundamentally enabling engineers to build the next generation of products faster and secure continuous performance gains in operations at scale for their customers.
We are doubling down on building foundational Physics AI for the industries that shape the physical world, such as aerospace, automotive, semiconductors, and energy.
Below are some of the published breakthroughs that this work rests on.
DEC 1, 2025
Going with the Speed of Sound: Pushing Neural Surrogates into Highly-turbulent Transonic Regimes
Existing aerospace datasets predominantly focus on 2D airfoils, neglecting these critical 3D phenomena. To address this gap, we present a new dataset of CFD simulations for 3D wings in the transonic regime. The dataset comprises volumetric and surface-level fields for around 30,000 samples with unique geometry and inflow conditions.
arXiv
NOV 25, 2025
Fluid Intelligence: A Forward Look on AI Foundation Models in Computational Fluid Dynamics
Driven by the advancement of GPUs and AI, the field of Computational Fluid Dynamics (CFD) is undergoing significant transformations. This paper bridges the gap between the machine learning and CFD communities by deconstructing industrial-scale CFD simulations into their core components.
arXiv
OCT 17, 2025
AB-UPT for Automotive and Aerospace Applications
In this technical report, we add two new datasets to the body of empirically evaluated use-cases of AB-UPT, combining high-quality data generation with state-of-the-art neural surrogates.
arXiv | Github
OCT 8, 2025
GyroSwin: 5D Surrogates for Gyrokinetic Plasma Turbulence Simulations
Nuclear fusion plays a pivotal role in the quest for reliable and sustainable energy production. A major roadblock to viable fusion power is understanding plasma turbulence, which significantly impairs plasma confinement, and is vital for next-generation reactor design.
arXiv | Github
FEB 23, 2025
AB-UPT
Anchored-Branched Universal Physics Transformer (AB-UPT) for aerodynamics CFD. Handles raw geometry without remeshing at 9M surface and 140M volume cells on a single GPU.
arXiv | Github
NOV 14, 2024
NeuralDEM
First end-to-end deep learning surrogate for large-scale multi-physics processes. Enables real-time simulation of industrial processes like fluidised bed reactors.
arXiv | Github
FEB 19, 2024
UPT: Universal Physics Transformer
A Framework For Efficiently Scaling Neural Operators across diverse spatio-temporal problems. Supports both grid and particle simulations.
arXiv | Github