Skip to main content
. 2026 Sep 9;28:e86453. doi: 10.2196/86453

Table 4.

Overall performance ranking of the selected large language models for the medical translation task from French to English, based on individual ranking results from 3 independent embedding models.

Model Size Embedding similarity ranking Overall rank


gte-multilingual-base bge-m3 nomic-embed-text-v1.5
Phi4 14B 1 1 1 1
Falcon3 10B 2 2 3 2
Mistral-small 24B 3 3 4 3
Meditron3 8B 5 4 2 4
Llama3.1 8B 4 5 5 5
Meditron3-Phi4 14B 6 6 6 6