Tag
This study evaluates four proprietary LLMs (GPT-4o, GPT-5.2, Claude Sonnet 4.5, DeepSeek) for specialized terminology translation from English to French across two domains, comparing prompting strategies. Results show Claude Sonnet 4.5 performs best, but LLMs cannot yet replace specialized corpora.