In-house development of ‘Ivy’, a platform based on ultra-large generative AI
CLEVI Co., Ltd. (Clevi, CEO Lee Hwan-ho) announced that it will launch the open beta service for ‘Ivy’, an ultra-large generative AI developed in-house using LLM (large language model) technology.
Canadian startup CLEVI has officially launched its in-house ultra-large generative AI model, ‘Ivy’, using LLM (large language model) technology. Until now, Canada had not achieved notable results in the AI market amid the dominance of OpenAI in the United States and DeepSeek in China. However, by officially launching its ultra-large generative AI, ‘Ivy’, CLEVI has demonstrated that Canada can also develop AI using its own technological capabilities.
Ivy offers powerful features, including △real-time learning and adaptation △multi-data processing △code interpreter △expert model generation and △agent support systems, allowing it to be used across a range of fields, from individual users to enterprises.
Over a period of more than two years, CLEVI CEO Lee Hwan-ho developed the company’s own model, starting with pre-training through an efficiently designed learning pipeline. According to the company, the model cost approximately KRW 1.5 billion to develop, while operating costs average approximately KRW 10 million per month.
CEO Lee emphasizes that “‘Ivy’ has high efficiency and high performance, enabling it to solve the issue often cited as a limitation of generative AI: ‘PoCs are easy, but completion is not.’” This allows companies to reduce labour costs and improve productivity by having AI perform repetitive and inefficient tasks on their behalf.
CLEVI stated, “To minimize AI development costs, we introduced innovative technology that efficiently manages the development process by applying automated, container-based learning node deployment technology,” adding, “We significantly reduced costs through the efficient design of the learning pipeline and optimization of the communication network between nodes.”
According to CLEVI, the ‘Ivy’ model is characterized by the development of a Mixture of models for efficient model training and operation. It was developed by creating small models and distilling and combining them into a model. Through multilingual training and embedding optimization, it supports not only English and Korean but also a variety of languages from around the world.
With U.S.-based OpenAI and China-based DeepSeek competing aggressively in AI technology, the domestic large-scale generative AI ‘Ivy’ has drawn industry attention 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), according to the explanation.
As concerns about personal information leaks and security have grown with the expanded use of AI Assistant, CLEVI has made the Ivy app platform available so that users can sign up with minimal personal information. In addition, public institutions and companies can install it on-premises to suit their user environments, which has been evaluated as addressing security concerns.
A CLEVI representative said, “Our company will not stop at the launch of Ivy. In the future, we plan to release an agent that directly intervenes, responds and participates in conversations between individual and group users.” The representative added, “We have also completed research and development on the AI-based proactive workflow agent ARI (Adaptive Reasoning Intelligence), as well as methods for precise object recognition and real-time control in areas such as video content production and medical image analysis, and are preparing to launch these services.”
Meanwhile, Ivy (‘Ivy’) is registered on the Google Play Store, and Ivy can also be downloaded and used through the company’s official website and the Ivy download page.
Choi Yong-seok, Donga.com reporter [email protected]
Source: The Dong-A Ilbo (reporter Choi Yong-seok) | https://www.donga.com/news/It/article/all/20250325/131275337/1