Portugal launches ‘Amália’ AI model as open-source alternative for Portuguese

Portugal will officially launch Amália, its first large language artificial intelligence model designed specifically for ‘European Portuguese’, on Wednesday, making the technology available as open-source software for government bodies, businesses, universities and researchers.

The launch, confirmed by Prime Minister Luís Montenegro’s office, marks a significant milestone in Portugal’s drive to develop sovereign AI capabilities tailored to the country’s language and culture, while reducing reliance on foreign technology.

Named Amália—short for Assistente Multimodal Automático de Linguagem com Inteligência Artificial (Automatic Multimodal Language Assistant with Artificial Intelligence, and alluding to iconic fado singer Amália Rodrigues)—the model is a large language model (LLM) capable of understanding, processing and generating natural language from vast datasets.

Originally unveiled by Mr Montenegro during the 2024 Web Summit in Lisbon, the project had been expected for several months after Education, Science and Innovation Minister Fernando Alexandre said earlier this year that development had produced “very positive results”.

Unlike commercial AI assistants such as ChatGPT, however, Amália is not being launched as a consumer-facing chatbot. Instead, it will serve as an open technological platform that public institutions, companies and researchers can build upon to create their own AI applications.

The government said the model, datasets and source code will all be released under an open-source licence, allowing organisations to reuse, improve and develop new services based on the technology.

Public sector rollout

The first deployments will focus on Portugal’s public administration, including education, defence, healthcare, culture and citizen services.

Among the initial applications planned are a virtual guide for Portuguese museums and monuments, an AI teaching assistant to help teachers prepare lessons, a digital citizen services assistant and decision-support tools for the Portuguese Navy.

The government says the project’s broader ambition is to create AI that better reflects the nuances of European Portuguese while also serving Portuguese-speaking countries across the wider Lusophone (Portuguese-speaking) world.

€5.5 million investment

Amália has received an initial €5.5 million in funding through Portugal’s PRR (Plan for Recovery and Resilience).

The largest allocations went to NOVA University Lisbon, which received €2.475 million, and Instituto Superior Técnico, which received €1 million. The Universities of Porto, Minho and Coimbra each received €375,000, while €900,000 was allocated directly to the Foundation for Science and Technology (FCT).

The project also benefits from Portugal’s investments in high-performance computing infrastructure, including the Deucalion and MareNostrum 5 supercomputers.

A test version of the model was completed in September 2025 and delivered to the FCT before being presented to the international scientific community at the PROPOR conference in Brazil earlier this year, where officials said it received a positive reception.

Built on European foundations

Although presented as Portugal’s national AI model, Amália was not developed entirely from scratch.

Researchers built upon the existing European EuroLLM-9B model before substantially expanding it with additional European Portuguese training data, increasing its capacity and context window, refining it with new Portuguese-language datasets, and developing enhanced safety, evaluation and performance optimisation systems.

The model has also been upgraded to process both text and images, giving it multimodal capabilities.

More than 60 researchers and students from a consortium including NOVA FCT, Instituto Superior Técnico, the Universities of Coimbra, Minho and Porto, and the Foundation for Science and Technology contributed to the project.

Funding for the next phase has already been secured through to the end of 2027 via Portugal’s Agency for Technological Reform of the State, and additional PRR funding.

Source: Executive Digest