Naver releases a new HyperCLOVA X model

Released a new HyperCLOVA X model — parameters cut to about 40%, operating costs improved by more than 50%, MMLU 79.6%, video understanding added

Why it matters

A turning point toward efficiency for a Korean lab

Source — Primary sources first; secondary reports are labelled “Press”

Officialnavercorp.comhttps://www.navercorp.com/media/pressReleasesDetail?seq=32288

Added in Week 35, 2026 · See that week

Same week

Google DeepMind researchers present SigLIP 2PaperxAI releases Grok 3 and Grok 3 mini in beta, plus DeepSearch, its first agentOfficialMTEB community presents MMTEBPaper

Nearby on the same axis

NSA — Native Sparse AttentionPaperNie et al. present LLaDAPaperDeepSeek-V3 Technical ReportPaperByte Latent TransformerPaper