South Korean telecom company KT has launched Token Factory, a business platform that gives companies access to 73 AI models through a single API, with automatic model selection and consolidated billing. Introduced in Seoul on October 27, the service was scheduled for its corporate launch on October 30.
The opening catalog mixes seven models served through KT’s domestic infrastructure — including its own Mideum K 3.0, Upstage’s Solar Pro 4, and Motif Technologies’ Motif 3 — with 66 global models, among them OpenAI’s GPT series, Anthropic’s Claude, and models from Mistral AI, DeepSeek, xAI, and Alibaba Cloud. KT has partnered with OpenRouter and plans to grow the catalog to more than 500 models by the end of 2026.
The centerpiece is the Auto Model Router, which chooses a model based on the question’s type and difficulty, the quality required, and the cost. Instead of naming a model, users set their preferred balance between quality and spending. KT claims prompt compression plus routing can cut token use by up to 25% — a company claim, not an independent measurement. A fallback switches failed requests to a pre-selected backup automatically.
Billing adapts KT’s telecom-style usage records: Token Detail Records log which models a customer uses and how many tokens they consume, underpinning one bill in Korean won. The Standard plan charges for actual token use with postpaid settlement, while Enterprise adds dedicated networks, private environments, and service-level agreements.
KT says the platform will not store customer data or use it for training, and screens inputs and outputs for sensitive and harmful content. Corporate management functions arrive by December, with access for individual developers planned for March 2027. Businesses comparing flagship assistants in the catalog can start with our ChatGPT vs Gemini comparison, alongside ChatGPT and Gemini.
Source: Superpower Daily
