Portugal has launched its first open-source national AI model, Amália, releasing its model, datasets and source code to help governments, researchers and businesses build sovereign European Portuguese AI applications.
Portugal has officially launched Amália, its first national large language AI model, as an open-source platform, making its model, datasets and source code available under an open-source licence for governments, businesses, universities and researchers to reuse, improve and build new AI applications.
Designed specifically for European Portuguese, Amália is part of Portugal’s push to develop sovereign AI capabilities tailored to its language and culture while reducing reliance on foreign AI technologies. Unlike consumer AI assistants such as ChatGPT, the model is intended as public digital infrastructure that organisations can adapt for their own use cases.
The first deployments will target Portugal’s public sector, including education, defence, healthcare, culture and citizen services. Planned applications include an AI teaching assistant, a virtual guide for Portuguese museums and monuments, a digital citizen services assistant, and decision-support tools for the Portuguese Navy.
The project received €5.5 million in initial funding through Portugal’s Recovery and Resilience Plan (PRR), with major allocations going to NOVA University Lisbon, Instituto Superior Técnico, the Universities of Porto, Minho and Coimbra, and the Foundation for Science and Technology (FCT). It also leverages the Deucalion and MareNostrum 5 supercomputers.
Rather than being developed entirely from scratch, Amália builds on the European EuroLLM-9B foundation model, which researchers expanded with European Portuguese datasets, increased model capacity and context window, enhanced safety and evaluation systems, and added multimodal text and image capabilities.
More than 60 researchers and students from Portuguese universities and research institutions contributed to the project. A test version was completed in September 2025 and presented at the PROPOR conference in Brazil. Funding has already been secured for the project’s next phase through 2027.
