HyperCLOVA X Surpasses 300,000 Downloads One Month After Open-Source Release

Number of Downloads Ranks in Top 0.03% of All Models
Visual Language Model 'SEED 3B' Records About 270,000 Downloads

Naver Cloud announced on May 26 that its three open-source artificial intelligence (AI) models, "HyperCLOVA X SEED," which were released on April 24, have surpassed 300,000 downloads in just over a month since their launch.

Naver Cloud made the three HyperCLOVA X SEED models available on April 24 through "Hugging Face," a community for sharing open-source AI models. The number of downloads for HyperCLOVA X SEED ranks within the top 0.03% of all models registered on Hugging Face.

is announcing at the press conference held on the 23rd of last month. Provided by Naver Cloud

Among the three, the visual language model (VLM) "HyperCLOVA X SEED 3B," which can process text, images, and video information together, was the most widely used, recording approximately 267,000 downloads.

HyperCLOVA X SEED 3B achieved higher average scores than similarly sized models from major U.S. and Chinese tech companies on nine benchmarks that evaluate visual information comprehension in both Korean and English. It also demonstrated a comparable accuracy rate to much larger overseas models with significantly more parameters.

The text models "HyperCLOVA X SEED 1.5B" and "HyperCLOVA X SEED 0.5B" also recorded about 35,000 downloads, and 29 derivative models have been created based on these two models.

Seong Nakho, Head of Hyperscale AI Technology at Naver Cloud, stated, "It is meaningful that the generative AI model developed by Team Naver 'from scratch' is gaining attention in the global open-source community." He added, "We hope that HyperCLOVA X will contribute to an AI ecosystem where models that understand various languages and cultures, including Korean, can coexist, thereby fostering greater diversity and leading to more innovative technologies and services."

Naver Cloud plans to introduce the HyperCLOVA X inference model currently under development through the conversational AI service "CLOVA X" in June. In addition, services utilizing the HyperCLOVA X speech model are also planned for future release.