30B-Class LLM Trained from Scratch
Released on Hugging Face; Free to Use
Fast Responses and Low Inference Costs

Krafton said on October 21 that it had released a preview version of its self-developed 30B-class large language model (LLM), "Raon-1 Mini."


Raon-1 Mini. Provided by Krafton.

Raon-1 Mini. Provided by Krafton.

View original image

The model is a preview released ahead of its official launch and is available on Hugging Face under the Apache 2.0 License. Anyone can download and use the model for free, and it can be used freely for research as well as for developing commercial services and products.


Raon-1 Mini Preview is a 30-billion-parameter LLM for which Krafton carried out the entire process, from model design to training from scratch. It supports English and Korean. Krafton designed a proprietary model architecture to achieve both high performance and inference efficiency. It lowered inference costs by combining a mixture-of-experts (MoE) structure, which activates only 4 billion of the model's 30 billion parameters as needed, with its own linear attention technology.


Krafton emphasized that Raon-1 Mini Preview delivers superior performance to overseas models of a similar size and comes close to large domestic models on some metrics. It highlighted the model's natural Korean conversation abilities and fast response times in particular. The model also excels at following instructions accurately, including those specifying formats and personas.


Krafton Releases Preview of Its Self-Developed LLM "Raon-1 Mini" View original image

Krafton plans to improve Raon-1 Mini's performance through updates. When the official version is released, the company will also publish a technical report detailing the model's training process and final performance.



"The release of Raon-1 Mini is the result of independently developing a competitive language model by drawing on Krafton's AI research and development capabilities, which it has built up over more than five years," said Lee Kangwook, Krafton's chief AI officer (CAIO). "We will expand the use of AI across a range of industries with a cost-effective LLM that delivers high performance while keeping costs down."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing