TildeOpen LLM: Europe’s Sovereign Multilingual AI

An open-source, foundational LLM (Large Language Model) for European languages – secure, adaptable, and ready for governments, institutions, and enterprises. 

Talk to us
Translate with TildeOpen

June 2024

Tilde wins Large AI Grand Challenge

September 2024

Access to LUMI
supercomputer obtained

March 2025

Model training
begins 

September 2025

Model goes live on Hugging Face

February 2026

TildeOpen goes live in

Tilde MT

Supported by:
AI_BOOST logo

Your language deserves better AI

Most AI models are built for the world’s major languages – and over 90% of LLM training data is in English. That means Baltic, Slavic, and other European languages are left behind, leading to lower accuracy, weaker cultural understanding, and limited access to high-quality AI tools.

That’s why we’ve developed TildeOpen LLM – an open-source foundational large language model with over 30 billion parameters, built to support all European languages. You can fine-tune it to your own needs and deploy it securely – locally or in the cloud – to build trustworthy AI that actually speaks your language. 

Find out more about TildeOpen LLM:
TildeOpen research team members sitting in the office
Billion parameters
25 +
Focus languages
28 +
GPU hours on LUMI
2 M

Why TildeOpen?

  • Customisable with your own data 
  • Secure and fully in your control 
  • Deployable on-premises or in the cloud
  • Integrates with existing systems and workflows 
  • Built as a foundation for advanced AI solutions
TildeOpen team member working with LUMI interface

The AI foundation you can trust

TildeOpen is more than a technological achievement. It’s an open-source foundation for custom AI, benefiting over 155 million Europeans.

Custom AI solutions for businesses and organisations

Adapt TildeOpen to your industry, data, and workflows — from virtual assistants to secure translation, speech tech, and more.

National language model development for governments

Build inclusive language models that serve public needs, promote digital sovereignty, and support all official EU languages.

Reliable performance across focus languages

TildeOpen consistently demonstrates strong linguistic accuracy and comprehension in public benchmarks

TildeOpen performs strongly on the MultiBLiMP benchmark, which measures a model’s ability to distinguish between grammatical and ungrammatical sentences. Lower error rates reflect stronger grammar modelling and more reliable text generation. View full benchmark results.

TildeOpen delivers higher efficiency in morphology-rich European languages thanks to a tokeniser and architecture designed specifically for them. Compared to LLaMA-3, it is 41% more efficient in Latvian, 37% in Lithuanian, 31% inFinnish, and 28% in Estonian and Polish, while also surpassing GPT and Mistral models. This translates to faster text generation performance in local deployments and consequently lower running costs for the same amount of data. View full benchmark results.

TildeOpen-30B achieves a state-of-the-art result on the Belebele reading comprehension benchmark, with an average accuracy of 84.7%. The model outperforms other locally deployable models such as Gemma-27B, ALIA-40B and EuroLLM-22B. View full benchmark results.

Powered by supercomputers, backed by Europe 

The development of TildeOpen is supported by the European Commission and powered by EuroHPC JU’s top-tier supercomputers – LUMI and Jupiter. By winning the Large AI Grand Challenge, we’ve been granted 2 million GPU hours on LUMI to execute this ambitious project.

LUMI supercomputer used for TildeOpenLLM

Contribute to a multilingual future

To build a strong multilingual LLM with over 30B parametrs, we’re looking for language data from across Europe. We welcome contributions from authors, publishers, state libraries, and other partners –  with flexible terms that work for you. 
Contact us
TildeOpen team of researchers

Get started on Hugging Face

Head over to Hugging Face to explore the TildeOpen-30b repository and access the full technical documentation.

TildeOpen on Hugging Face
Data providers that have already contributed to the project:

Our promise

Committing to open collaboration

Governments can leverage TildeOpen to create tailored language models that improve public service accessibility for all citizens.

Integrity and security

We’re continuously working towards minimising harmful or inaccurate content in TildeOpen, so it can be a trusted resource for diverse public use cases.

Open access

TildeOpen will be available for both commercial and non-commercial use under a permissive license, published in Hugging Face and ELRC-SHARE.

Knowledge sharing

We are committed to collaboration and sharing insights, inviting partners to work with us in advancing TildeOpen for the benefit of all.

Frequently asked questions

The TildeLM project aims to create a multilingual foundational large language model that focuses on underrepresented Baltic and Eastern European languages to promote digital equity and enhance access to advanced AI technologies for these communities.
This imbalance has efficiency and cost consequences. For instance, longer sequences are required to encode the same amount of information in lower-resourced languages compared to English, making models less efficient and more expensive to run. Additionally, the English-centricity of these models can introduce undesirable cultural biases. TildeLM will be trained to ensure equity for all supported languages.
The project targets Eastern European and Baltic languages such as Bulgarian, Croatian, Czech, Estonian, Finnish, Latvian, Lithuanian, Macedonian, Montenegrin, Polish, Serbian, Slovak, Slovene, and Ukrainian. The model will also support bigger languages such as English, French, German and Russian in balanced proportions to support translation and related multilingual tasks.
The LUMI (Large Unified Modern Infrastructure) supercomputer is the fifth fastest supercomputer globally and the fastest in Europe. It is part of the EuroHPC Joint Undertaking, a collaborative effort involving the European Union and European countries to create a world-class high-performance computing (HPC) ecosystem in Europe. The LUMI supercomputer is located in Kajaani, Finland.
The purpose of the Large AI Grand Challenge, funded by the European Commission, is to expand European AI frontiers by harnessing the potential of large-scale AI models. The participants in the competition were innovative startups and SMEs with the technical capacity to develop AI models that boost Europe’s competitiveness in Generative AI. The European Commission has announced the winners of the Large AI Grand Challenge. Four innovative AI companies from Europe, including Tilde, will share a prize of €1 million and 8 million computational hours to advance Europe’s leadership in AI development.

Tilde is a leading European language technology innovator and service provider with a mission to promote language diversity in the digital age. Tilde has over 150 employees in three offices located in Riga, Vilnius, and Tallinn. Tilde’s research team is comprised of nine PhDs and their research associates and has authored over 260 scientific publications. Over the years, Tilde has developed a vast R&D partnership network with leading EU research centres and universities and serves as a language technology research hub for the Baltic region.

 

Most recent research and development activities of Tilde are focused on foundational large language models (LLMs), fine-tuning of LLMs for downstream applications, and integration of instruction-tuned LLMs in natural language processing applications (e.g., machine translation, virtual assistants, retrieval-augmented generation systems, processing of spoken language, summarisation, etc.).

Build AI that speaks your language  

TildeOpen gives you the foundation to create secure and sovereign AI. Explore the model now or talk to us about tailoring it to your needs.