Tilde enters the global AI race with a European multilingual model and safeguards against disinformation

Latvian AI company Tilde has unveiled TildeOpen LLM, an open-source large language model trained on Europe’s flagship supercomputer LUMI.

RīGA, LATVIA, October 16, 2025 /EINPresswire.com/ -- With 30 billion parameters, TildeOpen outperforms global rivals in European languages and introduces built-in safeguards against Russian disinformation.
Developed by a 12-person team at Tilde Research Lab, TildeOpen delivers benchmark results that rival or surpass those of Big Tech while costing a fraction to develop. The model excels not only in major European languages but also in smaller ones that global models typically neglect.

Strong performance across European languages

Most commercial AI models are trained primarily on English, leading to poor results for smaller European languages and leaving over 200 million Europeans underserved. TildeOpen directly addresses this gap by covering 34 languages, including 24 official EU languages, plus Ukrainian, Norwegian, Icelandic, Turkish, and Balkan languages.
In public benchmarks, foundational model TildeOpen-30B sets a new state of the art on the Belebele reading comprehension benchmark, with an average accuracy of 84.7%, outperforming leading open models such as Gemma-27B, ALIA-40B, and EuroLLM-22B.
Thanks to a tokenizer and architecture adapted to morphology-rich languages, TildeOpen is also significantly more efficient: 41% better in Latvian, 37% in Lithuanian, 31% in Finnish, 28% in Estonian, and 28% in Polish compared to LLaMA-3, also surpassing GPT-4o and Mistral. This makes it a faster, more accurate, and more sustainable alternative for Europe’s diverse linguistic landscape.

Safeguarding European AI from disinformation

TildeOpen integrates dedicated safeguards to filter disinformation and propaganda, protecting the model from hostile influences. Recent investigations by VIGINUM, the Atlantic Council’s DFRLab, and GLOBSEC documented that the Pravda network (also known as Portal Kombat) have infiltrated Russian state-aligned disinformation into global AI models. To counter this, Tilde collaborated with media monitoring authorities to filter disinformation from training data, and applied topic-modeling to block politically sensitive content from Kremlin-controlled sources. Unlike global models hosted abroad, TildeOpen can be deployed locally or in trusted European cloud environments, ensuring compliance with EU data protection standards, the AI Act, and the forthcoming General-Purpose AI Code of Practice. By combining these safeguards with local deployment options in trusted European cloud environments, TildeOpen is trustworthy by design.

Delivering on Europe’s AI innovation promise

TildeOpen is among the first outcomes of the European Commission’s commitment, made in President von der Leyen’s 2023 State of the Union address, to unleash Europe’s supercomputing power for AI startups and SMEs. The Large AI Grand Challenge, launched in November 2023, awarded Tilde €250,000 in funding and 2 million GPU hours on LUMI. TildeOpen was trained using 1,5M GPU hours on 768 AMD MI250X GPUs.
Using remaining compute resources on LUMI, Tilde is working to extend TildeOpen’s context length from 8k to 64k tokens, enabling it to process entire books, legal documents, or technical manuals in one operation. The team is also preparing instruction-tuned versions aimed at specialized European tasks such as legal translation and e-government services.

Open and available to all

TildeOpen is released on Hugging Face (TildeAI/TildeOpen-30b) under a CC-BY-4.0 license, allowing full flexibility for researchers, universities, startups, public administrations, and companies. It can be fine-tuned for applications ranging from healthcare and education to media and government services, fueling innovation while safeguarding European values.

About Tilde

Tilde is a leading European AI-driven language technology company headquartered in Latvia. Through cutting-edge research, extensive multilingual datasets, and expert teams, Tilde advances language equality in the digital age and develops trustworthy AI for Europe’s languages.

Anna Paulina
Tilde
+371 25 402 002
anna.paulina@tilde.ai
Visit us on social media:
LinkedIn

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.