Monday, November 9, 2026

Motif Technologies raises investment from NVIDIA Corporation; to unveil 'Motif 4' in Q1 next year

Input
2026-11-09 19:14:54
Updated
2026-11-09 19:14:54
Motif Technologies logo. Provided by Motif Technologies
[Financial News] South Korean artificial intelligence (AI) company Motif Technologies has begun developing its next-generation large language model (LLM), 'Motif 4.' The company plans to use an investment round expected to include NVIDIA Corporation to strengthen the technological competitiveness of its AI models and accelerate commercialization.
According to industry sources on the 9th, Motif Technologies has completed the fundraising process for developing Motif 4 and begun full-scale model development. NVIDIA Corporation is expected to join the bridge round alongside existing investors. 
Motif 4 is the follow-up to the open-weight AI model 'Motif 3,' unveiled last August. The company plans to complete development and release the model weights in the first quarter of next year.
Motif 3 was developed through the Ministry of Science and ICT's Independent AI Foundation Model project. According to the company, last August it scored 47 on the overall intelligence index of Artificial Analysis, a global AI model evaluation organization, ranking 10th in the assessment of all AI models and fourth among open-weight models.
Motif Technologies said it developed Motif 3 in about five months using 768 graphics processing units (GPUs), applying proprietary technologies including its in-house-developed 'Group Differential Latent Attention (GDLA).' For its next model, the company plans to strengthen its competitiveness across the entire development process, from architecture design to post-training.
Motif Technologies CEO Lim Jeong-hwan said, "Creating differentiation at every stage, from architecture to post-training, is the only way to survive in the global frontier race. With Motif 4, we will demonstrate our ability to build a world-class global frontier model independently."
He added, "One regret with Motif 3 was that time and resource constraints kept us from doing enough post-training, including reinforcement learning (RL). With Motif 4, we will deliver a result that offers not only technical excellence but also usability and commercial viability."
[email protected] Jang Min-kwon Reporter