Portugal now has its own open-source AI model, and the launch feeds into Europe’s push for more homegrown AI infrastructure.
The government-backed model, called Amália, is built for European Portuguese and is designed as a foundation for public institutions, businesses, universities, and researchers to build AI tools. Rather than launch as a consumer chatbot, Amália will support applications in public services, education, culture, defence, and research while giving Portugal a local alternative to foreign AI systems.
The launch puts Portugal into a wider European race to develop AI systems that reflect local languages, public-sector needs, and regional control over critical technology.
Amália is built for Portugal’s version of Portuguese
Portugal launched Amália on Wednesday as its first open-source artificial intelligence model, Reuters reported. The model was developed by a consortium of Portuguese universities and research institutions with government support and €5.5 million in EU recovery funds, or about $6.26 million.
Prime Minister Luís Montenegro tied the launch to Europe’s broader push for AI independence. “Europe’s strategic autonomy is today, perhaps more than ever, tied to AI,” he said at the launch event, according to Reuters.
Amália takes its name from both Portuguese culture and the model’s technical role.
The Next Web said that Amália stands for Automatic Multimodal Language Assistant with Artificial Intelligence, while also nodding to Amália Rodrigues, the late fado singer closely tied to Portuguese identity.
Its language focus is central to the project.
European Portuguese refers to the form of Portuguese used in Portugal, which differs from Brazilian Portuguese in vocabulary, grammar, phrasing, and cultural references. That distinction matters because large commercial models are often trained on far more Brazilian Portuguese text than Portugal-specific language data.
For public services, education, museums, and other citizen-facing tools, those differences can affect tone, accuracy, and trust. A system built for Portugal’s own language patterns may be more useful than a larger model that treats Portuguese as one broad category.
Open source gives institutions more control
Amália is not meant to compete with ChatGPT as a consumer app. Instead, it is a foundation model that other organizations can adapt for their own AI services.
Amália, along with its training dataset and source code, has been released under an open-source license. The model ships with its weights, datasets, and code published openly, allowing users to inspect how it was trained, adapt it, and run it on their own hardware.
That openness is especially important for government and public-sector use. A model used in citizen services, defence, healthcare, or education needs more than convenience. It needs auditability, security controls, and the ability to adapt to the national language, policy, and data needs.
Portugal Resident noted that initial deployments will focus on public administration, including education, defence, healthcare, culture, and citizen services.
Planned applications include a virtual guide for Portuguese museums and monuments, an AI teaching assistant for lesson planning, a digital assistant for citizen services, and decision-support tools for the Portuguese Navy.
The planned applications show why Portugal is treating Amália less like a chatbot launch and more like a public digital infrastructure. Its real test will be whether government agencies, universities, and companies build useful services on top of it.
Portugal joins Europe’s sovereign AI race
According to Reuters, Portugal’s launch follows similar European efforts to reduce dependence on US AI providers such as OpenAI, Google, and Anthropic. France and Germany have already backed homegrown AI companies such as Mistral AI and Aleph Alpha as Europe looks for stronger regional alternatives.
Amália was not built entirely from scratch.
The model is based on EuroLLM-9B, a European foundation model, and was expanded by more than 60 researchers and students with European Portuguese datasets, a larger context window, stronger safety and evaluation systems, and multimodal capabilities for handling images as well as text.
The project also leans on Portugal’s high-performance computing investments. Reuters reported that Amália has access to the Deucalion and MareNostrum 5 supercomputers, providing researchers with the computing power needed to train and run large AI models.
Portugal Resident also said that the funding for the next phase has already been secured through the end of 2027. That gives the project more room to move from launch announcement to real-world adoption.
The next test is whether Amália becomes widely used outside the research community.
Publishing an open model is only the first step. The project will depend on public institutions, companies, universities, and developers to turn the model into applications useful for daily operations.
For Europe’s AI strategy, Amália is a small but specific example of what sovereignty can look like: open code, local language support, public-sector use cases, and infrastructure that does not depend entirely on foreign AI platforms.
Read more: Austria Urges EU to Explore Hosting Anthropic After US AI Access Restrictions