Introducing TowerLLM, Multilingual by design

Unbabel’s Generative AI model is the best performing machine translation on the market, enabling our customers to scale globally with lower costs and higher accuracy.

Proudly partnering with public and private agencies and organizations

TowerLLM captures the power of Generative AI and optimizes it for translation

  • Unparalleled Accuracy

    Delivers the most accurate translations, outperforming GPT-4o, GPT-3.5, Google, and DeepL across language pairs and domains.

  • Beyond Translation

    Excels at diverse translation tasks beyond traditional MT, including post-editing, named entity recognition, terminology consistency, and quality estimation.

  • Built on Quality

    Trained on billions of words of the highest quality translation data, curated using Unbabel’s Quality LLM, COMETKiwi, for superior translation performance.

The first multilingual LLM optimized for translation

Highly accurate translations, preserving meaning and nuance across 20 language pairs and various domains.

Automatically corrects Machine Translation output based on the AI Quality Estimation.

Identify and preserve named entities, such as person names, organizations, and locations, ensuring accuracy and consistency in translations.

By leveraging retrieval augmented generation (RAG), TowerLLM can retrieve relevant information from customer-specific data, such as glossaries, translation memories, and previously translated content, and incorporate it into the translation process.

language pairs for machine translation

Specific language-related tasks

Perceived accuracy achieved

Test TowerLLM Quality Now

Privacy at the forefront

Privacy at the forefront, no cost to performance

To build cutting-edge LLMs, Unbabel uses public and proprietary data. We safeguard sensitive information through meticulous anonymization before model training, ensuring data privacy throughout development and deployment.

Start Using TowerLLM today

Talk to your Account Manager about testing and implementing TowerLLM into your pipelines. It’s easy to get going via the Portal, API or your Unbabel integration.

To learn more, join our product and research team in a customers-only Office Hours.

Our Results

To benchmark TowerLLM performance against leading translation solutions like GPT-4o, GPT-3.5, Google and DeepL, we evaluated each model’s translation across 14 language pairs and four domains. We also assessed its prowess in multilingual reasoning and comprehension, by comparing TowerLLM’s performance to other Large Language Models. Read more about our research.

Language Pairs

TowerLLM surpasses Google, DeepL, GPT-3.5, and GPT-4o in translations from English to Portuguese, German, Spanish, French, Italian, and Korean.

Domains

TowerLLM significantly outperforms competitors in translations between English and German for Financial, Legal, Medical, and Technical domains. 

Comprehension & Reasoning

TowerLLM excels in multilingual benchmarks assessing language comprehension, common-sense reasoning, and contextual understanding, outperforming larger models.

Block background

Proudly built by Unbabel’s
Research & Development Team

Test TowerLLM Quality Now