In-house development of the ‘Ivy’ platform based on a foundation generative AI model
CLEVI Co., Ltd. (Clevi, CEO Lee Hwan-ho) announced that it would begin the open beta service for ‘Ivy’, a foundation generative AI developed in-house using LLM (large language model) technology.
South Korean start-up CLEVI has officially launched its own foundation generative AI model, ‘Ivy’, using LLM (large language model) technology. Until now, South Korea had not achieved notable results in the AI market, which has been dominated by Open AI in the United States and DeepSeek in China. However, the official launch of ‘Ivy’ by CLEVI, a South Korean start-up, has demonstrated that South Korea can also develop AI using its own technological capabilities.
Ivy provides powerful features including △real-time learning and adaptation △multi-data processing △code interpreter △expert model generation and △agent support systems, enabling it to be used in a wide range of fields, from individual users to enterprises.
Over a period of more than two years, CLEVI CEO Lee Hwan-ho developed the model from pre-training onwards through an efficiently designed training pipeline. According to the company, the model development cost approximately KRW 1.5 billion, while operating costs average around KRW 10 million per month.
CEO Lee emphasises that “‘Ivy’ has high efficiency and high performance, enabling it to solve the problem identified as a drawback of generative AI: ‘PoC is easy, but Complete is not.’” This enables companies to reduce labour costs and improve productivity by having AI perform repetitive and inefficient tasks on their behalf.
CLEVI emphasised that “to minimise AI development costs, we introduced innovative technology that efficiently manages the development process by applying automated, container-based training node deployment technology,” adding that “we significantly reduced costs through the efficient design of the training pipeline and optimisation of the communication network between nodes.”
According to CLEVI, a feature of the ‘Ivy’ model is the development of Mixture of models for efficient model training and operation. It was developed by creating small models, distilling them into and merging them with the model, and supports not only English and Korean but also various languages worldwide through multilingual training and embedding optimisation.
It is explained that, as US OpenAI and China’s DeepSeek compete fiercely to take the lead in AI technology, the domestic large-scale generative AI ‘Ivy’ has attracted attention in the industry by recording higher scores than the two existing models in various benchmark tests (△AIME 2024: 81% △MATH-500: 98.6 △SWE-BENCH: 51.1 △MMLU: 91.2).
As concerns over personal data leaks and security have grown with the wider use of AI Assistants, CLEVI has made the Ivy app platform available so that users can sign up with a minimum amount of personal information. In addition, public institutions and companies can install it on-premises to suit their user environments, which has been praised as a solution to security issues.
A CLEVI representative said, “Our company is not stopping at the launch of Ivy. We plan to unveil agents that will directly intervene, respond and participate in conversations between individual and group users in the future. We have also completed research and development into AI-based proactive workflow agent ARI (Adaptive Reasoning Intelligence), as well as precise object recognition and real-time control methods for video content creation and medical image analysis, and are preparing to launch these services.”
Meanwhile, Ivy (‘Ivy’) has been registered on the Google Play Store, and Ivy can also be downloaded and used from the company’s official website and the Ivy download page.
Choi Yong-seok, Donga.com reporter [email protected]
Source: Donga Ilbo (reporter Choi Yong-seok) | https://www.donga.com/news/It/article/all/20250325/131275337/1