📊 Full opportunity report: AMÁLIA · The Three Hard Questions. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Portugal’s national language model, AMÁLIA, is now operational and outperforms previous models on many benchmarks. However, fundamental questions about openness, native data, and objectives remain unresolved, raising concerns about its strategic value.
Portugal’s €5.5 million investment in the AMÁLIA language model has resulted in a functioning system that surpasses some benchmarks, but critical questions about its openness, native-language data, and strategic goals remain unanswered, raising concerns about the country’s AI direction.
AMÁLIA, a consortium project involving around 60 researchers from Portugal’s top institutions, was officially launched in October 2025. It is based on extending a multilingual European foundation model, EuroLLM, rather than training from scratch, contrasting with Italy’s Minerva approach. The model currently handles text only, with multimodal capabilities planned for future versions, and is accessible via the FCT’s IAedu platform to 450,000 users in higher education.
Technically, AMÁLIA has outperformed previous open models on European Portuguese benchmarks and beats Qwen 3-8B on most Portuguese-specific tests, though it still lags behind on certain key benchmarks like ALBA. The model’s training involved 107 billion tokens, with a modest 5.8 billion tokens from Portuguese web archives, representing roughly 5.5% of the total pre-training mixture. The supervised fine-tuning phase included 17-18% Portuguese data, but native-language emphasis remains limited.
While the technical progress is clear, public discourse, notably from researcher Duarte O.Carmo, has raised three critical questions: How open is ‘fully open’ really? How much native-language data is sufficient? And what should the model optimize for? These questions are central to evaluating the strategic value of AMÁLIA and similar models across Europe.
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.
Multilingual AI Translation Mastery: Building Accurate, Culturally Sensitive Language Tools and Global Communication Systems in 2026
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.
Portuguese Language Learning For Kids: Learn to Speak Portuguese with Ease! A Bilingual English-Portuguese Fun & Practical Activity Workbook for Children.
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 Data Preparation Guide: Fuel AI With Quality Data | Labeling Tools Explained | Human-in-the-Loop Best Practices | Prepare to Train Smarter | Annotate for Success | Annotation Drives Intelligence
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.
Large Language Models: The Hard Parts: Open Source AI Solutions for Common Pitfalls
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 Strategy and European Sovereign Models
The development of AMÁLIA highlights the broader challenge faced by European countries in building sovereign-language models that balance openness, native-language data, and strategic objectives. While technically promising, unresolved questions about openness and data sufficiency could influence Portugal’s future AI policies and its role within the European AI landscape. The questions raised by Duarte O.Carmo reflect ongoing debates about transparency, data sovereignty, and purpose-driven AI development, which are critical as the continent seeks to maintain autonomy from dominant US and Chinese models.
European Sovereign-Language Models and Strategic Challenges
Across Europe, multiple nations are investing in native-language large language models, such as Italy’s Minerva, Germany’s Aleph Alpha, and France’s Mistral. These efforts are part of a broader push for AI sovereignty amid concerns over dependence on US and Chinese technology. The European Union has also launched initiatives like the OpenEuroLLM consortium to foster collaborative development. Despite significant investments, many projects face similar structural questions about openness, native data, and objectives, which remain inadequately addressed publicly, complicating policy and strategic decisions.
Portugal’s AMÁLIA is a key case because of its publicly funded nature and the transparency surrounding its development, making it a focal point for evaluating how European nations are approaching these foundational issues. The ongoing debates reflect a broader uncertainty about how to best leverage native data and open models for national and continental AI competitiveness.
“The three questions about openness, native data, and objectives are not just technical concerns—they are strategic imperatives for European AI sovereignty.”
— Duarte O.Carmo
Unanswered Questions About Model Openness and Strategy
It remains unclear how open AMÁLIA will be in its final form, especially regarding access to training data and model weights. The precise criteria for native-language data sufficiency are still debated, and the ultimate strategic objectives—whether to prioritize academic, commercial, or governmental applications—are not yet publicly defined. Additionally, the impact of these choices on Portugal’s AI sovereignty and European integration remains an open question.
Next Steps for AMÁLIA and European Sovereign LLMs
The final version of AMÁLIA is expected in June 2026, which will likely clarify some of the current uncertainties. Over the next 12 to 24 months, Portugal and other European nations will need to address transparency around data and openness, evaluate model performance in real-world applications, and define clear strategic objectives. Public discussions, policy frameworks, and further technical developments are anticipated to shape the continent’s approach to sovereign-language AI models.
Key Questions
What makes AMÁLIA different from other European language models?
AMÁLIA is based on extending a multilingual foundation model rather than training from scratch, with a significant focus on Portuguese and a publicly funded, transparent development process.
What are the main concerns raised about AMÁLIA?
Key concerns include how open the model will be, whether it has enough native-language data, and what strategic goals it aims to serve—questions that impact Portugal’s AI sovereignty.
Why do these questions matter for Europe?
They reflect broader debates about maintaining technological independence, data sovereignty, and strategic autonomy amid global AI dominance by US and Chinese firms.
What are the next milestones for AMÁLIA?
The upcoming release of the final version in June 2026 will be critical for addressing transparency and strategic clarity. Over the following months, policy and technical evaluations will shape its future role.
Will AMÁLIA become commercially available?
Currently, it is accessible to academic users through the FCT platform. Future commercial deployment depends on final performance, openness, and strategic decisions by Portugal’s stakeholders.
Source: ThorstenMeyerAI.com
