Cafe24 expands LLM Router to 270 models with automatic selection

Cafe24 is promoting its LLM Router service, which connects multiple AI models through a single API and selects a model automatically based on task difficulty and cost, according to zdnet.co.kr. The company says the number of supported models has grown from about 120 at launch to about 270.

The automatic routing feature, named cafe24/auto, picks a model per request. Under the pay-as-you-go plan there is no monthly base fee and no contract, and failed calls are not billed. A real-time dashboard lets users check usage and costs.

The push comes as model routing and token cost management draw more attention. KT unveiled its own AI integration platform, Token Factory, on October 27 and began full service for enterprise customers on October 30. Token Factory initially provides 73 AI models through one API, with model selection, usage monitoring and billing, and KT plans to expand that to more than 500 models this month through a link with OpenRouter.

Also in the market is Reblue, which announced the first supply contract for its Continuum Enterprise AI infrastructure management solution on the 23rd of last month.

Cost control under scrutiny

According to OpenRouter, weekly token processing volume rose about 35-fold in a year, from 4.89 trillion in October last year to 172 trillion last month. In a survey of 2,050 IT operations and engineering experts worldwide, 55% said they cannot accurately control AI costs, and in the 2026 Splunk Observability Status report, 60% of respondents said unexpected billing spikes burden their organizations more than high baseline operating costs. IDC forecasts that by 2028, 70% of AI-leading companies will use advanced architectures that dynamically connect and manage diverse AI models.