🇮🇹 Models of Italy
2 models from Italy, best known first.
Italy's artificial intelligence sector expanded significantly in 2024, reaching a value of 1.2 billion euros. The country ranks highly in responsible AI metrics and is developing indigenous large language models to support its industrial base.
- Italian AI industry reached 1.2 billion euros in 2024
- Minerva 7B has 7 billion parameters trained on 1.5 trillion words
- Italy ranked 2nd in 2026 Global Index on Responsible AI
- AI4I institute based in Turin manages European InvestAI funds
- Leonardo supercomputer supports Minerva model development
Language models in Italy
The Italian artificial intelligence market recorded a substantial increase in economic value during 2024, totaling 1.2 billion euros. This growth reflects a broader integration of automated systems into manufacturing, finance, and healthcare sectors. Italy also demonstrates strong performance in ethical standards, placing second globally in the 2026 Global Index on Responsible AI. National efforts focus on translating academic research into practical industrial applications through dedicated institutes and European funding mechanisms.
History
Research activities began in the 1970s with work on automated reasoning and pattern recognition by scholars at the National Research Council. The field gained structure in 1988 with the creation of the Italian Association for Artificial Intelligence (AIxIA). This organization facilitated collaboration between universities and industry leaders. Key figures such as Roberto Navigli advanced multilingual natural language processing, laying groundwork for modern computational linguistics projects like BabelNet.
Labs and companies
Sapienza University of Rome leads model development through its NLP research group, producing the Minerva family of large language models. These systems are trained from scratch with a specific focus on the Italian language. The Istituto Italiano per l’Intelligenza Artificiale (AI4I) serves as a central hub for applied research and industrial partnership. It collaborates with CINECA and utilizes supercomputing resources like the Leonardo system to support training and innovation.
Languages
Minerva represents the first large language model family developed entirely from scratch with an emphasis on Italian linguistic nuances. The latest version, Minerva 7B, contains seven billion parameters and was trained on over 1.5 trillion words. Advanced instruction tuning techniques help reduce hallucinations and improve contextual accuracy. This approach aims to preserve cultural heritage while providing transparent, open-source alternatives to proprietary systems.
Policy and programmes
National initiatives are supported by the National Recovery and Resilience Plan to accelerate industrial adoption. Italy benefits from the European InvestAI programme, which mobilizes significant capital for artificial intelligence infrastructure. The EU plan includes a fund for building gigafactories equipped with high-performance computing chips. AI4I acts as the primary recipient of these funds in Turin, aiming to democratize access to advanced computational resources for smaller enterprises.
Today
In 2026, Italy scored 72.71 on the Global Index on Responsible AI, ranking second among 135 nations. This position places the country ahead of ninety-nine percent of assessed states and significantly above European averages. Industrial application growth in 2025 exceeded previous years in both value and sectoral reach. Ongoing investments aim to strengthen competitiveness in manufacturing, healthcare, and aerospace through public-private partnerships.
The facts in this overview come from the Wikipedia article "Artificial intelligence industry in Italy", written by its contributors and used under the CC BY-SA 4.0 licence. EveryLLM rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.