Reply announces a partnership with Mistral AI to develop sovereign and customizable artificial intelligence models

18 March 2026

Reply [EXM, STAR: REY] today announced a new partnership agreement with Mistral AI, a leader in innovative AI model development, aimed at accelerating the adoption of local, sovereign, secure and versatile generative AI solutions at scale.

At the core of the collaboration is a shared vision of sovereign AI, designed to enable organisations to adopt AI models while ensuring data control, protection of sensitive information, compliance with regulatory requirements and deployment on European infrastructures.

By combining Mistral AI’s open, high-performance AI models with Reply’s expertise in designing and customising Large Language Models using proprietary and domain-specific data, organisations in highly regulated sectors - such as public administration, defence, financial services, healthcare, telecommunications, and energy & utilities - can deploy tailored AI solutions that integrate seamlessly with existing systems. These solutions support the transformation of operational processes, enhance decision-making and deliver measurable business value, while ensuring the highest standards of security, data sovereignty and compliance.

A second core pillar of the Reply–Mistral AI collaboration is advanced AI model customisation, which enables the design and training of Large Language Models on proprietary and specialised datasets tailored to complex, data-intensive domains and ready for operational use.

Under this agreement, a project is currently under development with the Austrian Academy of Sciences, focused on the creation of a customised Large Language Model for the Greek language, spanning ancient, modern and papyrus texts. The model is designed to support researchers working with ancient Greek sources by providing advanced text search and text completion capabilities. It is trained on a highly specialised corpus that includes published ancient Greek literature, digitised papyri from multiple collections and selected modern Greek texts, curated from publicly available and scholarly sources.

Within this initiative, Reply and Mistral AI collaborate on the training and evaluation of the model to ensure accuracy, reliability and practical relevance. The project demonstrates how sovereign AI infrastructures and advanced model customisation capabilities can be effectively combined even in highly specialised and data-intensive domains.

“The integration of the Mistral AI ecosystem with Reply’s experience in developing AI solutions tailored to specific business process will enable organisations to deploy custom, secure and governable models, designed to ensure data sovereignty and data protection, that integrate seamlessly into existing operational workflows and scale reliably within enterprise architectures.” said Filippo Rizzante, CTO of Reply.

Mistral AI's Chief of Revenue Marjorie Janievicz said: "We are proud to partner with Reply. Together, we will help organizations deploy AI that meets their needs for performance, control, and customization."

Through this collaboration, Reply and Mistral AI provide a trusted and secure environment on European infrastructures, accelerating the adoption of advanced AI solutions while enabling organizations with stringent regulatory, privacy and data protection requirements to fully leverage generative AI.