Telecom

KT Joins Defense 5G and 6G Standardization, Ranks No. 2 Globally in AI Model Routing

TECHWORLD ·

[Photo: KT]

✦ AI Summary

KT will be the first Korean telecom company to participate in the 9th meeting of the NATO Wireless Communication Standardization Project (NWCSP).

The meeting will be held at KT Gwanghwamun Building West in Seoul from the 28th to the 2nd of next month, and will discuss ways to apply 5G and 6G to defense and the direction of standards development.

KT's in-house AI model routing technology, AutoModelRouter, ranked No. 2 in Acc-Cost Arena, a public benchmark within RouterArena.

KT is pushing to strengthen its global technological competitiveness by leveraging next-generation defense communications and multi-model AI operations technology. The company is stepping up efforts to expand its technology edge by presenting both its participation in defense communications standardization and its own AI technology achievements. KT's in-house AI model routing technology ranked No. 2 overall in a global public benchmark.

KT announced its participation in the 9th meeting of the NATO Wireless Communication Standardization Project (NWCSP). It is the first Korean telecom company to take part in NATO's discussions on 5G and 6G defense communications standardization. KT's qualification for participation is as a 5G and 6G standardization committee member.

The meeting is hosted by the Ministry of National Defense and the Telecommunications Technology Association (TTA). It will be held at KT Gwanghwamun Building West in Seoul from the 28th to the 2nd of next month. About 100 officials from domestic and overseas defense and government agencies, as well as telecom and standards experts, will attend.

The meeting will discuss ways to apply 5G and 6G to defense and the direction of developing standards for next-generation defense communications. KT will participate there as a standardization committee member.

NWCSP is a civil-military cooperation project in which NATO allied and partner government agencies and private-sector experts discuss the defense use and standardization of next-generation mobile communications technologies. In Korea, the Ministry of National Defense, the Korea Institute for Defense Analyses (KIDA), and TTA are participating. Starting with this meeting, KT will send 5G and 6G standardization experts.

At this meeting, KT will share its network deployment and operations experience accumulated through its defense 5G business. Based on its experience applying commercial 5G technology to the defense sector, KT will present its views on guidelines for introducing 5G- and 6G-based defense communications and on the direction of international standard development.

The meeting will share the current status of 5G and 6G technology development and defense applications by country. It will also review standard documents for satellite communications and wireless backhaul to apply next-generation communications technology to the defense sector. A key agenda item is ways to secure interoperability for connecting different communications technologies and systems across countries.

KT is moving to strengthen the competitiveness of its AI platform technology alongside next-generation communications. AutoModelRouter, KT's in-house AI model routing technology, took part in RouterArena, a public benchmark developed by researchers at Rice University in the U.S. RouterArena uses about 8,400 queries and evaluates response accuracy, cost efficiency, and robustness to input changes. In Acc-Cost Arena, which assesses accuracy and cost at the same time, AutoModelRouter ranked No. 2.

Research on AutoModelRouter has been accepted as an official paper for ICLR 2026, the International Conference on Machine Learning. The public leaderboard includes commercial technologies such as Microsoft's Azure Model Router.

AutoModelRouter is a technology that analyzes a user's task type, difficulty, and knowledge domain before automatically selecting the most suitable model from among multiple AI models. In the process, it considers both each model's response quality and usage cost. At the core of AutoModelRouter is reducing token usage costs while maintaining the quality level set by the enterprise.

This structure operates by assigning the appropriate AI model according to task difficulty and characteristics. For simple translation and information-checking tasks, it connects a cost-efficient model, while for specialized analysis and high-level reasoning tasks, it selects a high-performance model. From the user's perspective, it appears as a single AI service, but in the actual system structure, different models are assigned according to the nature of each request.

The same principle also applies to KT's Token Factory. Token Factory has functions for integrated management of multiple AI models and token usage environments. Within this system, AutoModelRouter plays the role of automatically selecting the model best suited to the task, thereby optimizing service quality and cost at the same time.

KT plans to continuously add new AI models to AutoModelRouter and further advance its multi-model operating environment. In defense, it also plans to strengthen civil-military technical cooperation with related organizations through its participation in NWCSP and expand its ability to respond to international standards for 5G and 6G defense communications.

Kim Won-tae, executive vice president and head of KT's Enterprise Public/Finance Business Division, said that applying commercial mobile communications technologies such as 5G and 6G to defense is a shared civil and defense task, and that the company will strengthen cooperation with related organizations for that purpose. He added that KT will also actively participate in discussions on the international standardization of next-generation defense communications technologies.

Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=407395

References

This article was produced with the help of an automated content generation algorithm.


Source: TECHWORLD

View original

This article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.