LG AI (EXAONE)

Korea's LG AI released its first VLM with 33B dense parameters.

A Qwen3-MoE-sized model from LG, rivaling its performance. It supports six languages (Korean, English, Spanish, German, Japanese, and Vietnamese), has a 256K context window with sliding window attention.

LG also continues to release new models. This iteration adds a dual-thinking mode, a 3:1 local:global attention and a focus on tool calling. However, it is released under a noncommercial license.

A series of reasoning models trained with SFT, followed by DPO and a GRPO variant as well as SEMPER during the RL phase. It performs surprisingly well given the low number of examples (10K) during the final RL phase.

English and Korean bilingual instruct model.