On Thursday, Samsung introduced the second iteration of its Gauss artificial intelligence (AI) model. Named Gauss2, this updated multimodal AI system is reported to offer enhancements in performance, efficiency, and versatility for application integration. The launch occurred during the online Samsung Developer Conference Korea 2024 (SDC24 Korea), where the company emphasized its commitment to leveraging the power of AI in software development and research initiatives.
Samsung Gauss2 AI Model Introduced
In a recent announcement, Samsung detailed the advancements in its Gauss AI model, which first debuted last year. The latest version incorporates several key upgrades, with notable enhancements in its multimodal capabilities, allowing the model to effectively manage data sets from various sources.
Gauss2 is now capable of supporting between 9 to 14 languages, in addition to various programming languages. Upgraded performance is reported in handling language, code, and image processing tasks.
The Samsung Gauss2 is offered in three distinct variants based on parameters: Compact, Balanced, and Supreme. The Compact model is designed for efficiency, ideal for environments with limited computational power. The Balanced model strikes a middle ground, optimizing both performance and efficiency, while the Supreme model is equipped to tackle high-demand computing tasks using Mixture of Experts (MoE) technology.
Samsung asserts that the Balanced and Supreme variations outperform “leading open-source generative AI models” in language-oriented tasks in both English and Korean, as well as in coding challenges. The company also highlights reduced wait times, quicker processing speeds, and enhanced task management compared to open-source alternatives.
Currently, the Samsung Gauss2 is utilized within the company’s Device eXperience (DX) division and various international research institutions. One of its primary applications is as an internal coding assistant, referred to as code.i, which supports software development endeavors, with 60 percent of the DX division reportedly engaging with this tool. Additionally, the technology is employed by call center personnel to categorize and summarize customer interactions.
While Gauss2 is presently an internal asset, Samsung plans to incorporate it into its forthcoming products. The company is optimistic that the AI model will enhance the personalization features of its existing applications.
