EXAONE 4.5: LG's First Open-Weight Vision-Language Model
arxiv.orgAI ResearchApr 9, 2026, 5:51 PM
EXAONE 4.5 (LG AI Research): first open-weight VLM from LG. Visual encoder integrated into EXAONE 4.0; native multimodal pretraining. 256K context, 6 languages. SOTA in document understanding and Korean contextual reasoning at similar scale. Open-weights on HuggingFace.
6Apr 13, 2026, 12:54 PM