This submission is a prompt-based terminology and segment injection strategy for terminology translation across domains. We evaluate the technique on Basque to Spanish, English to Polish, and Traditional Chinese to English for WMT 2026 Terminology Shared Task 1 & 2. We demonstrate that injecting bilingual terminology and bi-text examples into recent Large Language Models’ (LLM) prompts improves both overall translation quality and terminology accuracy. Our results show that strong instruction-following ability allows a system to adapt to terminology and style constraints without the need for fine-tuning.

The session will provide an overview of some of the new Machine Translation metrics available on the market, analyze if and how these new metrics correlate at a segment level to the results of Adequacy and Fluency Human Assessments, and how they compare against TER scores and Levenshtein Distance – two of our currently preferred metrics – as well as against each of the other. The information in this session will help to get a better understanding of their strengths and weaknesses and make informed decisions when it comes to forecasting MT production.