K-EXAONE

Training compute

1.5×10²⁴ FLOP

Parameters

236B

Published

Dec 31, 2025

K-EXAONE is an AI model developed by LG AI Research (South Korea), first published in December 2025. It works in the language domain, on tasks such as language modeling/generation, chat, question answering and 3 more.

Training it took an estimated 1.5×10²⁴ FLOP of compute. The model has 236,000,000,000 parameters. It was trained on roughly 11T datapoints. Training ran on 512 NVIDIA B200 GPUs for about 3.2K hours.

Access: API access. Its weights are not openly released. Epoch AI rates the confidence of this record as confident.

Full record

Organization: LG AI Research
Country of organization: South Korea
Domain: Language
Task: Language modeling/generation, Chat, Question answering, Code generation, Mathematical reasoning, Instruction interpretation
Training compute: 1.5×10²⁴ FLOP
Parameters: 236,000,000,000
Dataset size: 11T
Training hardware: NVIDIA B200 GPUs
Chips used: 512
Training time: 3,240 h
Model accessibility: API access
Open weights: No
Epoch confidence: Confident