Dnotitia Unveils ‘DNA 2.0’: A Korean-Language Foundation Model Built for Agentic AI

SEOUL, SOUTH KOREA, July 26, 2025 /EINPresswire.com/ -- Dnotitia Inc. (Dnotitia), a fast-rising innovator in long-term memory AI and semiconductor-integrated solutions, has announced the launch of its high-performance large language model (LLM), ‘DNA 2.0’. Moving beyond traditional text generation, the model is designed to enable agentic AI – AI that can autonomously integrate with external tools and perform complex tasks. As agentic AI becomes a core technology driving business automation and operational efficiency, ‘DNA 2.0’ represents a significant step forward in real-world AI applications.

The model features advanced tool-calling capabilities that allow it to understand user instructions and perform tasks such as search, summarization, and computation. A Korean-optimized post-processing framework refines command structures and response formats to deliver more accurate and consistent outputs.

Dnotitia has applied a Model Context Protocol(MCP) – compatible architecture, enabling the automatic conversation of natural language commands into internal specifications and seamless API integration with external tools. This underpins essential agent functionalities such as Korean-language RAG (Retrieval-Augmented Generation) and persona configuration.

‘DNA 2.0’ is available in a wide range of parameter sizes, from compact 0.6B models to large-scale 235B MoE(Mixture of Experts) configurations. The MoE structure enhances computational efficiency by activating only the relevant expert modules based on context. The 14B version is released as open-source, while the smaller models are optimized for on-device deployment.

In parallel, Dnotitia has launched its Korean LLM Leaderboard, providing an objective benchmark for evaluating LLMs on Korean-language performance across both commercial and open-source models. Built on an internally developed evaluation dataset tailored for Korean, the leaderboard enables side-by-side comparisons under consistent conditions. On this benchmark, the newly released ‘DNA 2.0’ demonstrated superior performance compared to most Korean-language models of similar size.

“DNA 2.0 is not just a generative AI model,” said Moo-Kyoung Chung, CEO of Dnotitia. “It’s a practical agentic AI designed for real-world applications in Korean. It presents a viable alternative for achieving high performance with limited resources and lays the groundwork for building sovereign AI tailored to national languages, legal systems, and cultural contents.”

Following the release of ‘DNA 1.0’ last year, Dnotitia has now introduced three LLMs to date, including ‘DNA-R1’, Korea’s first logic reasoning model specialized for Korean. The company has also open-sourced ‘QLLM-INFER’, a benchmark tool for evaluating AI quantization algorithms, and Smoothie Qwen, a multilingual post-processing tool that reduces unintended language outputs. Through these open-source initiatives, Dnotitia is contributing to the advancement and wider adoption of generative AI technologies in both domestic and international communities.

Dnotitia Communications
Dnotitia, Inc.
email us here
Visit us on social media:
LinkedIn
Facebook
YouTube

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.