📊 Full opportunity report: ALIA. The Spanish answer. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Spain has announced ALIA, its largest publicly funded AI project, featuring a 40-billion-parameter multilingual model trained on 9.37 trillion tokens. While operationally below Llama 2, ALIA emphasizes Spanish-language coverage and widespread adoption, highlighting strategic positioning over performance.
Spain has officially launched ALIA, its largest public-funded multilingual large language model (LLM), trained on 9.37 trillion tokens across 35 languages, with a focus on Spanish. This project is part of the broader discussion on hyperscaler investments and AI market dynamics. The project, led by the Barcelona Supercomputing Center and funded with over €240 million, aims to position Spain as a key player in multilingual AI adoption within Europe.
The ALIA-40B model, released under the Apache License 2.0 on HuggingFace on April 22, 2025, is part of Spain’s national AI strategy and is designed to serve the Spanish-speaking world and co-official languages. It was trained on MareNostrum 5’s 4,480 NVIDIA H100 GPU-accelerated partition, with the project structured around three layers: political leadership, technical coordination, and originating projects such as AINA and ILENIA.
Benchmark results show ALIA’s performance is below that of Llama 2 at comparable scales, with Llama 2 achieving 66% accuracy on XNLI_en and 93-94% on SQuAD_en, compared to ALIA’s 51.77% and 81.53%, respectively. Understanding hyperscaler capex trends can shed light on the strategic positioning of models like ALIA. This indicates a structural capability gap, aligning with the project’s strategic emphasis on multilingual coverage and adoption rather than top-tier performance.
Josep M. Martorell, ALIA’s project leader, stated the goal is not to be the most performant LLM globally but to maximize adoption within the Spanish-speaking world, reflecting a Position 3 strategic profile focused on operational credibility and regional influence rather than performance supremacy.
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.
Spanish language AI chatbot
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.
large language model training dataset
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.
NVIDIA H100 GPU for AI
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications of ALIA for European AI Strategies
ALIA represents the most ambitious national AI project in Europe in terms of public investment, scale, and scope. Its emphasis on multilingual coverage and Spanish-language oversampling aligns with Spain’s strategic goal to foster regional AI adoption and influence. The project exemplifies a shift from performance-centric to operational and strategic positioning, potentially shaping future European AI initiatives and collaborations.
Despite its lower benchmark performance compared to Llama 2, ALIA’s open-source release and AESIA validation bolster its credibility as a transparent, regionally focused AI model. This approach may influence other European nations to prioritize multilingual and regional relevance over raw performance metrics.
Spain’s National AI Investment and Strategic Positioning
Spain’s ALIA project follows a series of European national AI initiatives, including Portugal’s AMÁLIA, Italy’s Minerva, and France’s Mistral. The project is part of Spain’s broader €150 million investment under the Spanish AI Strategy 2024, supplementing €90 million for MareNostrum 5 upgrades. It operates within a framework of public funding and institutional coordination led by the Secretary of State for Digitalisation and Artificial Intelligence and the Barcelona Supercomputing Center.
Training on MareNostrum 5’s high-performance infrastructure and the open-source release of Salamandra-40B mark a significant step for Spain’s AI ambitions, emphasizing multilingual capabilities and regional adoption. The project’s focus on Spanish and co-official languages aims to foster domestic and regional AI ecosystems, contrasting with other European projects that prioritize performance or commercial deployment.
“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
Operational Performance and Strategic Positioning Clarity
While benchmark results confirm ALIA’s performance is below Llama 2, it remains unclear how this will affect its adoption and impact within Spain and Europe. The long-term success of the project’s strategic focus on regional influence over raw performance is still to be seen, and further operational data will clarify its effectiveness.
Next Steps for ALIA Deployment and Evaluation
Future developments include broader deployment of ALIA across Spanish government agencies and industry, ongoing benchmarking, and potential updates to improve performance. These efforts are connected to the ongoing discussion about the $725 billion hyperscaler capex. Monitoring how the model is adopted and integrated into regional AI ecosystems will be key to assessing its strategic success.
Key Questions
What is the main goal of Spain’s ALIA project?
The primary goal is to promote widespread adoption of a multilingual AI model within the Spanish-speaking world, emphasizing regional influence over top-tier benchmark performance.
How does ALIA compare to other European AI models?
Benchmark results show ALIA’s performance is below that of Llama 2 at similar scales, but it offers broader multilingual coverage and regional focus, which are central to its strategic positioning.
What are the technical specifications of ALIA?
ALIA is a 40-billion-parameter model trained on 9.37 trillion tokens across 35 languages, leveraging MareNostrum 5’s NVIDIA H100 GPU infrastructure, and released under Apache License 2.0.
What is the significance of ALIA’s open-source release?
The open-source release under Apache 2.0 enhances transparency and collaboration, supporting Spain’s goal of regional AI development and fostering regional ecosystems.
What are the next milestones for ALIA?
Next steps include broader deployment, ongoing benchmarking, and potential updates aimed at improving performance and expanding regional adoption.
Source: ThorstenMeyerAI.com
