Baek Kyung-hoon, Head of LG AI Research. Photo courtesy of LG.

 

 

[Alpha Biz= Kim Jisun] LG AI Research is set to unveil the next-generation version of its artificial intelligence (AI) model, Exaone 4.0, in July, integrating large language model (LLM) capabilities with advanced reasoning functions. According to industry sources on June 24, development is in its final stages alongside Exaone Path 2.0, a pathology-focused AI model for cancer diagnosis.



Exaone 4.0 builds upon Exaone 3.5, released in December 2023, and integrates the reasoning AI engine Exaone Deep, introduced in March. The new version is expected to deliver significantly improved performance by combining the strengths of both models, offering high-reliability answers to complex problems in specialized fields such as science, mathematics, biology, and chemistry. Exaone 3.5 was previously capable of processing documents as long as 100 A4 pages in a single prompt.



Meanwhile, Exaone Path is an AI model designed for analyzing histopathological images. The upcoming Exaone Path 2.0 will aim to drastically reduce the time and cost of cancer diagnostics by shortening genetic test processes that typically take up to two weeks.



LG AI Research plans to release both models as open source, allowing external researchers and developers to inspect and validate the models directly.

 

 

 

[ⓒ AlphaBIZ. 무단전재-재배포 금지]