| Mobile Web

GPU-backed companies speed next-generation LLM and action AI development

Startups that received government support for graphics processing units are speeding up development of AI models, including diffusion-based transformer large language models, Korean-focused document parsing models and action AI agents, officials and industry sources said on Wednesday. Trillionlabs built its Trida-7B model using Nvidia H200 GPUs, while Posicube and Inhance are developing Korean VLMs and action models. The Science Ministry is allocating and monitoring GPUs funded by a supplementary budget.