📊 Full opportunity report: AMÁLIA · The Three Hard Questions. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Portugal’s AMÁLIA, a €5.5 million European Portuguese language model, is operational but prompts three key questions about its openness, native data sufficiency, and optimization goals. These issues highlight broader challenges for Europe’s sovereign AI efforts.
Portugal’s €5.5 million AMÁLIA language model is now operational and accessible to academic users, but it faces three fundamental questions about its openness, native language data, and strategic objectives, raising broader concerns for Europe’s sovereign AI initiatives.
AMÁLIA is a consortium project involving approximately 60 researchers across Portugal’s leading institutions, including NOVA, IST, and IT. The model, announced in December 2024 and released in October 2025, is based on a continuation of the EuroLLM multilingual foundation, rather than training from scratch. It handles Portuguese text, with multimodal capabilities planned for future releases, and is currently used by 450,000 academic users through the FCT’s IAedu platform.
Technically, AMÁLIA outperforms previous open models on Portuguese benchmarks and beats Qwen 3-8B on most Portuguese tasks, although it still trails Qwen on certain benchmarks like ALBA. The model’s training included 107 billion tokens, with a small portion (around 5.8 billion) from Portugal’s web archive Arquivo.pt, representing roughly 5.5% of the extended pre-training data. The supervised fine-tuning phase involved about 17-18% Portuguese data, with no dedicated native-language pre-training emphasis. The final version is scheduled for release in June 2026, with ongoing assessments of its capabilities and limitations.
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.
APRENDA GitLab CI/CD: Implemente DevOps com Deploys Automatizados e Feedback Contínuo (Infraestrutura & Automação Brasil) (Portuguese Edition)
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.
The Complete Google Agent ADK Blueprint: Build 150+ Multimodal AI Agents with Google's Agent Development Kit, Gemini and Google Cloud (The Complete AI Blueprint Book 11)
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.
Training Data for Machine Learning: Human Supervision from Annotation to Data Science
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.
HRH Spanish ESP Language Silicone Keyboard Cover Protector for MacBook Pro 13 inch 2020 (Model A2289 / A2251 / A2338 M1 Chip) and for Pro 16" 2019 (Model A2141),European Version
1)【European Layout only】COMPATIBLE WITH: (1) Pro 13-Inch 2020 with M1 processor (model: A2338)(2) Pro 13-Inch Early 2020 released…
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.
Structural Challenges for Europe’s Sovereign LLMs
The development of AMÁLIA exemplifies broader issues facing European AI sovereignty efforts. The three critical questions—how open is ‘fully open,’ how much native-language data suffices, and what should models optimize for—are not only technical but also strategic. These questions influence national policies, research priorities, and Europe’s ability to develop autonomous, competitive language models. The ongoing debates and unresolved issues could shape the future landscape of European AI independence and innovation.
European Sovereign LLMs and Strategic Uncertainty
Across Europe, multiple countries and initiatives—Italy’s Minerva, Germany’s Aleph Alpha, France’s Mistral, and others—are investing in sovereign language models. These efforts are driven by a desire for technological independence and control over AI capabilities. However, most projects are still in early stages, and the discourse tends to focus on individual model launches rather than the structural patterns and strategic choices that underpin these efforts. The case of AMÁLIA highlights the common challenges: balancing openness with proprietary concerns, determining the necessary native-language data volume, and defining clear objectives for model optimization.
Public analysis, including Duarte O.Carmo’s critique, emphasizes that these questions are often left unaddressed, risking strategic missteps and suboptimal investments. The Portuguese project is thus a microcosm of a larger European debate on how to build sustainable, sovereign AI infrastructure amid global competition and technological uncertainties.
“The three questions—openness, native data sufficiency, and optimization goals—are fundamental to understanding the strategic landscape of European sovereign LLMs.”
— Duarte O.Carmo
Unresolved Strategic and Technical Questions
It remains unclear how European projects will resolve the questions of openness, native-language data sufficiency, and optimization focus. The final version of AMÁLIA, scheduled for June 2026, may address some gaps, but current discussions do not yet provide definitive answers. The impact of these uncertainties on policy and investment strategies is still being evaluated.
Next Milestones and Ongoing Evaluations
The key next step is the release of AMÁLIA’s final version in June 2026, which will include further testing and possible enhancements. Researchers and policymakers will closely monitor its performance, openness, and strategic alignment. Additionally, European initiatives will continue to grapple with the three core questions, shaping the continent’s AI sovereignty trajectory over the next 12 to 24 months.
Key Questions
What are the main challenges facing AMÁLIA’s development?
The main challenges include defining what ‘fully open’ means in practice, determining how much native Portuguese data is enough for effective modeling, and setting clear priorities for what the model should optimize for—whether accuracy, safety, or strategic independence.
How does AMÁLIA compare to other European models?
AMÁLIA outperforms previous open models on Portuguese benchmarks and beats Qwen 3-8B on most Portuguese tasks, but it still trails Qwen on some benchmarks like ALBA. Its technical approach, based on continuation of a multilingual foundation, differs from models trained from scratch.
Why are these questions about openness and native data important?
They are crucial because they influence transparency, control over data, and strategic autonomy. How open a model is affects collaboration and trust, while native data volume impacts linguistic accuracy and cultural relevance, shaping Europe’s AI independence.
What will happen after the final version of AMÁLIA is released?
Researchers and policymakers will evaluate its performance, address remaining gaps, and decide on future development directions. The broader European AI community will also continue debating how to best answer these foundational questions to ensure sustainable sovereignty.
Source: ThorstenMeyerAI.com
