AI · SEMICONDUCTOR · FACT CHECK

SKD WIRE

EN WIRE · 2026-10-27 17:15

KT launches 'Token Factory' integrating 500 AI models, cutting token usage by up to 25%

Mostly True

KT launches 'Token Factory' integrating 500 AI models, cutting token usage by up to 25%
Image source: 조사 출처

KT has officially launched "Token Factory," a platform that unifies access to 500 AI models while reducing token consumption by up to 25 percent, according to the company's official newsroom.

원문 주장 (KR)
KT, AI 모델 500개 통합 '토큰팩토리' 출시…토큰 최대 25% 절감

One gateway to 500 AI models

Token Factory is designed to give users a single point of access to a large pool of AI models — 500 in total — rather than contracting with each provider separately. By routing requests through its own optimized infrastructure, KT says the platform can cut token usage by as much as 25 percent, lowering costs for enterprises building on large language models (LLMs).

Positioning in the enterprise AI market

The launch places KT, one of Korea's major telecom carriers, squarely in the enterprise AI infrastructure business, where token efficiency translates directly into cost savings for customers running AI workloads at scale. Details on pricing, supported model vendors, and availability timelines beyond the official announcement remain limited.

The claim that KT integrated 500 AI models into Token Factory with token savings of up to 25 percent matches the company's own announcement, and with no contradicting evidence on record, the assessment lands as Mostly True.

Sources — primary documents (8)
  1. https://tokenfactory.kt.com/coming-soon
  2. https://ax.kt.com/ax/solution/token-factory
  3. https://news.einfomax.co.kr/news/articleView.html?idxno=4441365
  4. https://www.ajunews.com/view/20261027130838820
  5. https://www.etnews.com/20261027000311
  6. https://www.hankyung.com/article/202610276766g
  7. https://ax.kt.com/api/v1/files/product/ebef1076-7c59-43f7-acb3-e40edff772e9.png
  8. https://cdn.aitimes.com/news/photo/202610/216709_220848_1238.jpg

KR: /news/20261027-d7824a · 판정: 대체로 사실