📊 Full opportunity report: AMÁLIA · The Three Hard Questions. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Portugal’s state-funded AMÁLIA large language model is now operational, outperforming some benchmarks. However, key questions about its openness, native data, and objectives remain unanswered, highlighting broader issues in European sovereign-LLM efforts.
Portugal’s €5.5 million investment in the AMÁLIA large language model has resulted in a functioning system that outperforms previous open models on Portuguese benchmarks, but fundamental questions about its openness, native data, and strategic goals remain unanswered.
The AMÁLIA project involves around 60 researchers from Portugal’s leading institutions, including NOVA, IST, and IT. It was announced in December 2024, with the base version completed by September 30, 2025, and publicly launched on October 1, 2025. The model is currently accessible to 450,000 academic users through the FCT’s IAedu platform, with knowledge cut off at the end of 2023. The project is structured as a continuation of the EuroLLM multilingual foundation, rather than training from scratch, contrasting with models like Italy’s Minerva.
Technically, AMÁLIA has demonstrated superior performance on Portuguese benchmarks, outperforming previous open models and beating Qwen 3-8B on most Portuguese-specific tasks, though it still trails on some benchmarks like ALBA. The model is still in development, with final adjustments expected before the June 2026 release of the final version.
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.
large language model development kit
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 AI training dataset
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 benchmarking tools
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.
European language AI platform
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 European Sovereign-Language AI Strategies
The development of AMÁLIA exemplifies a broader shift in European AI policy, emphasizing national sovereignty and language-specific models. However, the project highlights persistent uncertainties about how open these models truly are, how much native-language data is sufficient, and what strategic objectives they should pursue. These questions influence not only Portugal’s AI landscape but also the future of European independence in AI technology, potentially affecting data governance, transparency, and regional competitiveness.
European Sovereign-Language Models and Structural Challenges
Across Europe, several countries are investing in sovereign-language large language models, including Italy’s Minerva, Germany’s Aleph Alpha, France’s Mistral, and initiatives like OpenEuroLLM and AI Sweden. These efforts are motivated by desires for technological independence, data sovereignty, and linguistic preservation. Yet, most models are still in progress, and the community has yet to address core questions about openness, native data sufficiency, and strategic goals transparently. Portugal’s AMÁLIA is a key case because of its public funding and national scope, making these questions particularly relevant at the policy level.
“The three questions—how open is ‘fully open,’ how much native-language data is enough, and what should we be optimizing for—are central to evaluating the true progress of European sovereign models.”
— Duarte O.Carmo
Unanswered Questions About Openness and Strategy
It remains unclear how open AMÁLIA truly is, particularly regarding access to training data and model weights. The strategic goals—whether to prioritize linguistic preservation, commercial competitiveness, or technological independence—are still under debate. Additionally, the sufficiency of native Portuguese data and how it influences performance across diverse tasks have not been fully clarified. These uncertainties are compounded by the evolving nature of the project and the lack of transparent benchmarks on some key aspects.
Next Milestones for AMÁLIA and European LLMs
The final version of AMÁLIA is scheduled for release in June 2026, which will likely clarify many current uncertainties. Over the next 12-24 months, the project team is expected to publish more detailed benchmarks, openness policies, and strategic objectives. Wider European initiatives will also continue to grapple with these questions, potentially leading to more transparent standards for sovereign-language models and their deployment. Policymakers and researchers will be watching closely for signs of progress or persistent gaps.
Key Questions
What makes AMÁLIA different from other European LLMs?
AMÁLIA is a publicly funded project involving Portugal’s top research institutions, structured as a continuation of a multilingual foundation rather than training from scratch, with the goal of supporting Portuguese language tasks.
What are the main concerns about AMÁLIA’s openness?
It is still unclear how accessible the model weights and training data are, which affects transparency and potential for wider use or auditing.
Why are the three questions raised by Duarte O.Carmo important?
They address core issues of transparency, data sufficiency, and strategic purpose, which determine how effective and independent European models can become.
Will the final version of AMÁLIA address these questions?
It is expected that the June 2026 release will clarify some of these issues, but ongoing transparency and policy debates are likely to continue.
How does AMÁLIA compare to models like Italy’s Minerva?
While Minerva was trained from scratch on Italian data, AMÁLIA builds on a multilingual foundation, which influences its technical approach and strategic implications.
Source: ThorstenMeyerAI.com
