# High-precision image-text understanding

Heron-NVILA-Lite-33B is a vision-language model based on the NVILA-Lite architecture, specifically trained for Japanese, and supports multimodal tasks in both Japanese and English.