A Homegrown Genius Emerges

Imagine if Italy built its own version of ChatGPT — one that not only spoke fluent Italian but also truly understood Italian culture, context, and priorities.

That’s exactly what researchers at Sapienza University of Rome have done. They’ve launched Minerva 7B, a large language model (LLM) trained entirely in Italy using open data and public compute infrastructure.

What makes it such a big deal? Well, most advanced AI tools today — such as ChatGPT are generally products of U.S. tech giants, and they are built by using datasets that we do not have full access to, or that we cannot verify. Minerva, on the other hand, is a game-changer due to its full transparency, local development, and open source nature.

Just think of it as creating your own engine instead of using someone else’s.

Handmade from the Ground Up. No Inheritance.

Most European AI tools are based on rehashed models (publicly available ones, like OpenAI’s GPT or Meta’s LLaMA), and the main differentiator between Minerva 7B and the rest is that the former is built from scratch.

The majority of the tokens were learned from the datasets (in the tech world, tokens are like words broken down into smaller parts) that are Italian and English public texts — the authorship of the entire world of books and Wikipedia, government, etc., has now been transferred to these datasets for AI training.

This is also the thing that makes Minerva different from other AI when it comes to Italian — not only the language part of it, but how it thinks and interacts with you. It’s trained with a European mindset.

It’s Not Just Interlocutory — It’s Safe and Smart

Minerva 7B doesn’t just chatter away like other fact-hurled chatbots — it’s been instruction-tuned, which means it’s trained to follow prompts properly, reject harmful queries, and provide more thoughtful, coherent answers.

Minerva was definitely not made in someone’s garage.

It combined CINECA’s Leonardo, one of the top five fastest supercomputers in Europe, running 128 energy-efficient GPUs to achieve the training of the model within a span of only a few weeks. So Italy is not only in the AI race, but it is outrunning the competition.

Investing in such a different type of infrastructure is exactly what enables the countries to be the ones steering the trajectory of their AI future instead of depending on Silicon Valley.

Open Entirety for Anyone — From Devs to Educators

Minerva 7B isn’t locked behind a paywall or a tech monopoly. It aligns with an open science movement of a wider scope: the code, the training data, and the outcomes are all publicly accessible for anyone to review, use, and enhance. Consequently, developers are free to try out new ideas; researchers can do their experiments, and governments can even do the auditing.

This is the kind of thing that inspires people’s trust in AI and not just the usage of it.

Besides, Minerva has this ITA-Bench testing instrument that is designed to score the model based on how well it performs tasks like reasoning, math, and language.

An Overview of Europe’s AI Future

This isn’t just a “cool university project.” It’s part of a wider European strategy to create sovereign AI systems — meaning models developed within Europe, for Europe, under European laws and ethics.

With AI now being deeply tied to education, public services, healthcare, and even law, Europe doesn’t want to depend only on the U.S. or China for critical infrastructure.

Italy’s PNRR (National Recovery Plan) and the broader InvestEU initiative are now backing projects like Minerva as the future backbone of public-sector AI.

What People Are Saying

AI enthusiasts online are already exploring Minerva’s abilities. On Reddit and Hugging Face, people are testing its responses and comparing them to models like Mistral and LLaMA 2.

While some note that it’s still early days and more polish is needed, the potential is clear: a local, ethical, and highly capable AI assistant with a strong focus on trust and performance.

Why It Matters

• For citizens: AI trained on local languages means fewer misunderstandings, more accurate public services, and cultural nuance.
• For developers: Open-source access allows more innovation and fewer barriers to entry.
• For policy makers: Transparent, ethical AI helps align with data regulations like GDPR.
• For Europe: It’s about not falling behind and building an AI future on its own terms.

What’s Next?

The team behind Minerva plans to keep improving the model and may launch larger versions in the future. These could become foundational tools in European schools, hospitals, courts, and beyond.

Imagine AI that doesn’t just understand your words, but your world.

Final Thought

In a world dominated by American tech giants, Minerva 7B is a quiet but powerful shift — a local answer to a global question.

It’s not just Italy flexing its AI muscles. It’s Europe laying the bricks for a future where AI is open, ethical, and homegrown.

And that’s a future worth building.

Looking to build a high-performing remote tech team?

Check out MyNextDeveloper, a platform where you can find the top 3% of AI engineers who are deeply passionate about innovation. Our on-demand, dedicated, and thorough software talent solutions provide a comprehensive solution for all your software requirements.

Visit our website to explore how we can assist you in assembling your perfect team.