K-EXAONE 2.0 Technical Report

2026-08-05

AI Summary

This work proposes a multilingual Mixture-of-Experts (MoE) model based on the K-EXAONE architecture, featuring 750 billion total parameters (approximately 37 billion activated) and supporting a 256K-token context window across ten languages. To enhance comprehensive capabilities in reasoning, long-context understanding, agent-based coding, and culturally aligned safety—while avoiding the prohibitive cost of training from scratch—the model leverages a combined strategy of continual pretraining, difficulty-focused intermediate training, and post-training alignment. Notably, it incorporates Korean sociocultural context into its safety alignment mechanism for the first time. Released under the Apache 2.0 license, the model achieves over threefold capacity improvement and consistently outperforms its predecessor across nine evaluation categories, demonstrating particular strength in long-context retrieval, agent coding, and multilingual safety.

Abstract

This technical report presents K-EXAONE 2.0, an open-weight multilingual foundation model developed by LG AI Research as a step in our effort toward global frontier-scale foundation models. Rather than training from scratch, we upcycle K-EXAONE and expand its architecture, yielding a Mixture-of-Experts (MoE) model with 750B total parameters and approximately 37B activated per token---more than three times the capacity of its predecessor. K-EXAONE 2.0 supports context lengths of up to 256K tokens and expands multilingual coverage from six to ten languages. Its training pipeline combines continual pre-training, difficulty-focused mid-training, and post-training to strengthen reasoning, agentic coding, multilingual capability, and safety grounded in Korean sociocultural contexts. Across nine evaluation categories selected to reflect the conditions of practical use, K-EXAONE 2.0 improves over K-EXAONE and remains competitive with open-weight models, showing its largest gains in agentic coding and long-context understanding and its clearest strengths in long-context retrieval and safety. Released under the Apache 2.0 license, K-EXAONE 2.0 enables the wider AI ecosystem to evaluate, deploy, adapt, and build upon it, while marking the beginning---rather than the endpoint---of our challenge toward the global frontier.

Problem

• foundation model
• multilingual
• Mixture-of-Experts
• long-context understanding
• agentic coding

Innovation

• Mixture-of-Experts
• upcycling
• long-context understanding
• agentic coding
• multilingual foundation model