Naver Cloud announced on the 26th that the three models of the open-source AI "HyperCLOVA X Seed," released on the 24th of last month, have surpassed 300,000 downloads in just over a month.
According to Naver Cloud, this figure corresponds to about 0.03% of the models registered in the global artificial intelligence (AI) model sharing community "Hugging Face." The company noted, "The influence of HyperCLOVA X is expanding globally."
Naver Cloud reported that "HyperCLOVA X Seed 3B," a visual language model capable of processing text, images, and video information together, recorded approximately 267,000 downloads, making it the most popular among the three models.
The company stated that, in comparing the average scores of nine benchmarks that evaluate visual information understanding ability related to Korean and English, HyperCLOVA X Seed 3B scored higher than similar-sized models from American and Chinese big tech firms. It has demonstrated competitive performance by showing answer rates comparable to that of large-scale foreign models with significantly more parameters.
"HyperCLOVA X Seed 1.5B" and "HyperCLOVA X Seed 0.5B," both text models, also recorded about 35,000 downloads.
Naver Cloud plans to introduce the under-development HyperCLOVA X inference model through the interactive AI service "CLOVA X" in June. Additionally, they announced plans to sequentially showcase services utilizing the HyperCLOVA X voice model, expanding HyperCLOVA X's multi-modality from text to images and video, and further into voice.