📊 Full opportunity report: AMÁLIA · The Three Hard Questions. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Portugal’s AMÁLIA LLM, funded with €5.5 million, is now operational and outperforms many models on Portuguese benchmarks. However, key structural questions remain about its openness, native data sufficiency, and optimization goals, impacting national AI policy.
Portugal’s €5.5 million AI project, AMÁLIA, is now operational, with the model available to hundreds of academic users and outperforming previous open models on Portuguese benchmarks. However, critical structural questions about its openness, native data, and goals remain unanswered, raising concerns about the broader European sovereign-LLM movement.
AMÁLIA is a consortium project involving approximately 60 researchers from Portugal’s leading research institutions, including NOVA, IST, and IT. The project was announced in December 2024, with the base version completed by September 2025, and officially launched on October 1, 2025. The model is trained as a continuation of the EuroLLM multilingual foundation, not from scratch, and handles text-only tasks with plans for multimodal capabilities in future versions.
Technical evaluations show AMÁLIA outperforms previous open models on European Portuguese benchmarks and surpasses models like Qwen 3-8B on most Portuguese-specific tasks, although it still trails Qwen on some benchmarks like ALBA. The training data includes approximately 107 billion tokens, with about 5.8 billion tokens from Portuguese web archives, representing roughly 5.5% of the total pre-training mixture.
Despite these advances, questions about the model’s openness, the adequacy of native-language data, and the strategic goals guiding its development have been raised by critics such as Duarte O.Carmo, who emphasizes the need for transparency and clear objectives in national AI efforts.
AMÁLIA
The three hard
questions.
Portugal spent €5.5M to build a European Portuguese LLM. The base version is operational, the benchmarks beat Qwen 3-8B on most pt-PT tasks. So why are the most important questions still unanswered?
Last month, Duarte O.Carmo published the sharpest public analysis of AMÁLIA — Portugal’s state-funded European Portuguese large language model. He prefaces his critique with the necessary diplomatic apparatus before doing what almost nobody else in the European-sovereign-LLM discourse has been willing to do publicly: asking hard questions about whether the work, as released, actually does what it set out to do. This piece is a structural extension of his analysis. The AMÁLIA case study exposes three hard questions every national LLM effort needs to answer publicly — and the broader European sovereign-LLM movement has been operating without explicit answers to any of them.
Three questions every national LLM effort needs to answer publicly.
Duarte O.Carmo’s framing maps cleanly onto the structural argument. Each question lands specifically in AMÁLIA — and the broader European sovereign-LLM movement has been operating without explicit answers to any of them.
The three questions form a structural feedback loop. Q3 (optimization target) determines Q2 (data volume needed) which conditions Q1 (openness sufficient for community contribution). The European sovereign-LLM movement collectively benefits from these questions becoming standard methodology disclosure, not exceptional critique.
Portuguese language AI language model
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
107 billion tokens. 5.8 billion clearly pt-PT.
The structurally tractable question with a structurally surprising answer. For a model whose entire stated purpose is European Portuguese prioritization, the native-language share of extended pre-training is 5.5%. The implications cascade into every other question.
multilingual large language model
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Olmo standard. AMÁLIA’s current state.
Allen Institute for AI’s Olmo project defines what “fully open” operationally requires. Olmo doesn’t lead frontier benchmarks. That’s not the point. The point is to be the structural reference for openness. AMÁLIA’s “fully open source” claim should track to the operational standard.
AI model training dataset
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Four strategic positions. AMÁLIA between two and three.
Approximately €100M+ in publicly disclosed European sovereign-LLM funding across the major initiatives. The structural question every project faces: what is the actual competitive position you’re staking? Four options — none mutually exclusive — but each requiring different commitments.
AI model evaluation tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Three standards. For AMÁLIA and the movement.
The structural critique generalizes beyond AMÁLIA. Italy, France, Germany, Switzerland, the OpenEuroLLM consortium, and every subsequent national project benefit from public discourse holding national LLM efforts to operational standards on openness, data accounting, and strategic positioning.
The European sovereign-AI agenda is a serious strategic project that deserves serious public discourse. O.Carmo’s analysis is what serious public discourse looks like. Appropriately diplomatic. Structurally rigorous. Willing to ask the hard questions in public when the public investment justifies it. More of this is needed — across every European sovereign-LLM project, not just AMÁLIA.
Implications for Portugal’s AI Sovereignty and Policy
The development of AMÁLIA highlights the broader challenge facing European nations in balancing open AI development with strategic control. The unresolved questions about openness, native data, and goals directly influence Portugal’s AI sovereignty, its ability to compete with global models, and the transparency of public investments in AI. Addressing these issues publicly could set a precedent for responsible national AI strategies across Europe, ensuring models serve local needs while maintaining openness and accountability.European Sovereign-LLM Efforts and Structural Challenges
Across Europe, multiple countries and initiatives, including Italy’s Minerva, Germany’s Aleph Alpha, and France’s Mistral, are developing sovereign-language models with similar funding and strategic ambitions. These efforts are often characterized by questions of how open models should be, how much native-language data is enough, and what their primary objectives should be. The European movement is still grappling with these fundamental issues, often launching models without explicit answers to these core questions. Portugal’s AMÁLIA exemplifies this pattern, with its public funding and national scope making the questions more urgent and publicly relevant.“AMÁLIA is an impressive piece of work, but it raises fundamental questions about openness and strategic direction that remain unaddressed.”
— Duarte O.Carmo
Unanswered Questions About AMÁLIA’s Strategic and Technical Foundations
It remains unclear how open the final version of AMÁLIA will be, whether the native Portuguese data used is sufficient for future improvements, and what the ultimate strategic goals of the project are beyond benchmarking success. It is also uncertain how the Portuguese government and research institutions will address these questions publicly as the project progresses toward its final release in June 2026.
Next Milestones and Public Clarifications Expected by June 2026
The final version of AMÁLIA is scheduled for release in June 2026, and it is expected that researchers and policymakers will address the current questions about openness, native data sufficiency, and strategic objectives. Additionally, the project team may release more detailed documentation or open-source components to clarify these issues. Monitoring how Portugal’s government and research institutions communicate these priorities will be key to understanding the model’s future role and European AI policy directions.
Key Questions
What are the main concerns about AMÁLIA’s openness?
Critics question whether the model will be fully open, given the strategic importance of transparency and control over AI development, especially in a national context.
Is the native Portuguese data used in AMÁLIA sufficient?
While the model includes about 5.8 billion tokens from Portuguese web archives, some experts argue this may not be enough for comprehensive language understanding or future improvements.
What are the strategic goals behind AMÁLIA?
The official goal is to develop a high-performance Portuguese language model for academic and public use, but critics urge clarification on whether it aims for commercial competitiveness, sovereignty, or both.
Will the model be open source?
It is not yet confirmed whether AMÁLIA will be fully open source, though the project’s emphasis on openness remains a key question for the European sovereign-LLM movement.
What happens after the final release in June 2026?
Researchers and policymakers will likely evaluate the model’s capabilities, transparency, and strategic fit, potentially leading to policy adjustments and further development based on the unresolved questions.
Source: ThorstenMeyerAI.com
