en.Wedoany.com Reported - South Korea's Cafe24 recently launched a large language model (LLM) router that can connect over 120 generative AI models and automatically select the most suitable model for tasks such as coding, translation, and creation based on user requests.
As the "multihoming" phenomenon of generative AI models like ChatGPT, Claude, Gemini, and DeepSeek spreads, users face an increasing burden of having to compare and select models themselves. Cafe24's LLM router integrates these models through a single application programming interface (API), aiming to reduce users' decision-making burden.
Service operators can set which models to use based on criteria such as task type, cost, response speed, and model performance. The router also supports automatic switching to a pre-specified alternative model when a particular model fails, ensuring continuity of task execution.
The generative AI market has seen a "paradox of choice," where an increase in options can lead to decision delays or lower satisfaction after selection. Each AI has different strengths, and users alternate between multiple services depending on the task purpose—such as translation, search, or coding—but the process of comparing models and repeatedly entering the same questions lengthens decision-making time. Cafe24 launched the LLM router precisely to address this issue, allowing users to focus on the task itself rather than struggling with which model to choose.
According to the "2025 Survey on the Actual State of Value-Added Communications Business" by South Korea's Ministry of Science and ICT, 53.2% of generative AI users use more than two services simultaneously, indicating that multihoming has become widespread. Overseas, services that connect and mediate AI models have already formed a market. For example, OpenRouter, founded in 2023, provides over 400 generative AI models from more than 70 AI companies through a single API. It is reported to process over 100 trillion tokens per month and has more than 10 million users worldwide.
Cafe24 is entering South Korea's domestic generative AI mediation market with the LLM router as its entry point. Its goal is to provide an environment that automatically allocates models based on task purpose, cost, speed, and performance. Through automatic model selection and automatic switching in case of failure, it aims to improve the operational efficiency and stability of generative AI services, helping users focus on the tasks they need to perform.
