What is Minerva AI?
This listing covers the Minerva language-model project from Sapienza University of Rome, rather than unrelated products with the same name. The models were trained from scratch on Italian and English data. The project now also includes ChatMinerva, developed with Babelscape.
What can you use it for?
ChatMinerva supports conversations about uploaded images, scanned pages, reports and technical documents, and offers live web access. The instruction-tuned 7B model is designed for tasks including conversation, summarization and translation. These are different ways to use the project: the hosted assistant and downloadable models have separate requirements.
Access and running costs
Published Minerva models are available through Sapienza NLP on Hugging Face. The Minerva-3B-base-v1.0 model card lists an Apache 2.0 license and access conditions. Running a model yourself requires computing resources; downloadable weights do not mean that hosting, integration or the ChatMinerva service is cost-free. Check the terms for the exact model or service you plan to use.
What should you check before adopting it?
The 3B base model is a foundation model, not an aligned chat assistant. Its model card warns about inaccurate answers, bias and unsuitable content. Evaluate the specific version on your own tasks, and review hosting and data-handling terms before uploading confidential material. Italian development alone does not establish GDPR compliance for every deployment.
