📊 Full opportunity report: ALIA. The Spanish answer. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Spain’s ALIA-40B AI model, trained on 9.37 trillion tokens across 35 languages, is the largest publicly funded European project of its kind. It emphasizes multilingual coverage and Spanish oversampling but shows performance below Llama 2 benchmarks, confirming a capability gap.
Spain’s ALIA-40B, a 40-billion-parameter multilingual AI model, has been officially released under an open-source license, marking the country’s most ambitious public AI project to date. Developed by the Barcelona Supercomputing Center and funded with over €240 million, ALIA aims to serve the Spanish-speaking world and demonstrate Spain’s strategic AI capabilities.
The ALIA project, led by the Spanish government through the Secretary of State for Digitalisation and Artificial Intelligence, involves training on 9.37 trillion tokens across 35 European languages and 92 programming languages. For more on European AI investments and strategic questions. It was trained on MareNostrum 5’s GPU-accelerated infrastructure, utilizing 4,480 NVIDIA H100 GPUs, and released under the Apache License 2.0 on HuggingFace on April 22, 2025.
Official documentation from the Barcelona Supercomputing Center confirms that ALIA-40B is designed to prioritize Spanish and co-official languages, aligning with Spain’s strategic goal to promote multilingual AI adoption within its borders. The model’s benchmark performance against Llama 2 shows significant gaps—51.77% on XNLI in English versus Llama 2’s 66%, and 81.53% on SQuAD in English versus Llama 2’s 93-94%—indicating a structural capability gap.
According to Josep M. Martorell, ALIA’s project leader, the main goal is to maximize adoption within the Spanish-speaking world rather than competing for top benchmark scores, framing ALIA as a Position 3 strategic initiative focused on operational relevance over raw performance.
ALIA.
The Spanish
answer.
€240M+ Spanish public funding · ALIA-40B + Salamandra family · 9.37T tokens · 35 European languages + 92 programming languages · MareNostrum 5 · Apache 2.0 release. The largest publicly funded European national-AI project by cumulative scope — and the empirical test case for the Position 1 vs Position 3 strategic-positioning argument.
This is the tenth standalone essay in the European sovereign-LLM track and the third Tier 2 expansion piece. ALIA is Spain’s institutional answer — the largest EU member state by GDP not yet documented in the track. The project markets itself as Position 1 + Position 2 simultaneously — “Europe’s first public multilingual foundational model.” The benchmark evidence (ALIA-40B 51.77% XNLI_en vs Llama 2 66%) confirms the structural capability gap from Finding 1 of the synthesis essay. The Position 3 framing — Martorell’s “most widely adopted in the Spanish-speaking world” — is operationally honest. €90M MareNostrum 5 upgrade + €150M company integration = €240M+ cumulative scope. Apache 2.0 open-source release + AESIA validation + co-official languages oversampling. Both can be true at once. The Spanish public discourse would benefit from explicit Position 3 strategic positioning.
Six models. Apache 2.0.
The ALIA family operates as a tiered model portfolio. ALIA-40B is the flagship at 40 billion parameters; the Salamandra family scales down to 7B, 2B and instruct-tuned variants; mRoBERTa provides the foundational multilingual baseline. All released under Apache License 2.0 on April 22, 2025 at the HispanIA 2040 event — “Public Code, Public Money” approach.
multilingual
MN5 LLM
edge
target
instruct
encoder
multilingual AI language model
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Four official. Oversampled by factor of 2.
ALIA’s distinctive multilingual coverage strategy. The four co-official Spanish languages are oversampled by factor of 2 in the training corpus — structurally distinct from Apertus’s broad 1,811-language coverage approach. The strategy targets deep coverage of Spanish co-official languages rather than maximum language breadth.
open-source AI models for Spanish
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
ALIA-40B vs Llama 2. 14-point gap.
The empirical evidence Finding 1 of the synthesis essay needed. ALIA-40B at 40 billion parameters with €240M+ public funding and 8+ months MareNostrum 5 training achieves performance below Llama 2 — a 2023 frontier model released approximately 18 months before ALIA-40B. The capability gap is real and consistent with six of seven prior national-project answers documented in the track.
NVIDIA H100 GPU for AI training
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Two pilots. Public administration deployment.
The operational deployment targets that validate the Position 3 + Position 4 framing. Public administration deployment is the structurally credible Position 3 + Position 4 strategic positioning — captive demand from Spanish public institutions where Spanish-language specialization is operationally distinctive.
The work is real across the Spanish ALIA case. €240M+ public funding committed. 40B parameter from-scratch model trained on 9.37 trillion tokens. Salamandra family released under Apache 2.0. AESIA validation aligned with EU AI Act transparency standards. Two pilot applications shipped — Tax Agency chatbot and primary care medicine heart failure diagnosis. The Position 1 framing is operationally misleading. ALIA-40B performance below Llama 2 confirms the structural capability gap. The Position 3 framing is operationally honest — Spanish-speaking world adoption, co-official languages oversampling, public administration deployment. Both can be true at once. The Spanish public discourse would benefit from explicit Position 3 strategic positioning.
AI model training infrastructure
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications of ALIA’s Strategic Positioning and Performance
ALIA’s development underscores Spain’s commitment to establishing a sovereign AI infrastructure, emphasizing multilingual coverage and regional language support. While its benchmark performance lags behind leading models like Llama 2, its strategic focus on Spanish language adoption and open-source transparency positions it as a credible national effort aligned with European sovereignty goals. This approach highlights the tension between operational ambition and technical performance, influencing future AI policy and development strategies within Europe.
Background and Strategic Framework of Spain’s ALIA Initiative
The ALIA project, initiated by Spain’s government and coordinated through the Barcelona Supercomputing Center, is part of a broader €240 million public investment in AI infrastructure, including MareNostrum 5 upgrades. It follows a series of national AI efforts across Europe, such as Portugal’s AMÁLIA, Italy’s Minerva, and the pan-European OpenEuroLLM, but is the largest publicly funded project in terms of scope and scale within Spain.
Training on MareNostrum 5’s GPU infrastructure and using a multilingual dataset of over 9 trillion tokens, ALIA aims to serve as a national answer to Europe’s sovereign AI questions. Its design emphasizes multilingual capabilities, especially Spanish, with the intent to promote widespread adoption and operational utility rather than top-tier benchmark performance.
“The goal is not to be the best-performing LLM in the world, but the most widely adopted in the Spanish-speaking world.”
— Josep M. Martorell
Unconfirmed Aspects of ALIA’s Performance and Strategic Goals
It remains unclear how ALIA’s performance will evolve with further development and fine-tuning. The model’s benchmark results suggest a structural capability gap compared to top models like Llama 2, but the operational impact and real-world adoption levels are still to be assessed. Additionally, the long-term strategic positioning—whether ALIA will shift toward higher performance or remain focused on multilingual operational utility—is still under discussion.
Future Developments and Strategic Adjustments for ALIA
Next steps include ongoing benchmarking, fine-tuning, and potential updates to improve performance. The project team plans to expand multilingual capabilities and integrate ALIA into various applications across Spain. Monitoring adoption rates and operational effectiveness will inform whether ALIA maintains its Position 3 focus or shifts toward a more performance-oriented strategy.
Key Questions
What is the main goal of Spain’s ALIA project?
The primary goal is to promote widespread adoption of AI in the Spanish-speaking world, emphasizing multilingual coverage and operational utility over benchmark performance.
How does ALIA compare to other European AI models?
Benchmark results indicate ALIA lags behind models like Llama 2 in key NLP tasks, reflecting a focus on multilingual support rather than raw performance.
What are the strategic implications of ALIA’s development?
It demonstrates Spain’s commitment to sovereignty and regional language support, positioning itself as a Position 3 initiative that prioritizes operational relevance over top benchmark scores.
Will ALIA improve in performance over time?
Future updates and fine-tuning are planned, but whether performance will catch up with leading models remains uncertain.
Why is open-source release important for ALIA?
Open-source licensing under Apache 2.0 facilitates transparency, collaboration, and wider adoption within the Spanish-speaking and European AI communities.
Source: ThorstenMeyerAI.com
