K-EXAONE

K-EXAONE ranks first on WMT24++.

LG AI Research Released 31 Dec 2025 Family K-EXAONE The lab's model card

Benchmarks with a score

Ranked first, out of five or more models

In the top three

LG AI Research Dec 2025

Open weights Flagship

No Epoch Capabilities Index score yet

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Benchmark	Score	Rank	Source
WMT24++	90.5%	#1 of 12	Model card
IFEval	89.7%	#27 of 126	Model card
MMLU-Pro	83.8%	#41 of 151	Model card
KoBALT	61.8%	#1 of 3	Model card
MMMLU	85.7%	#28 of 62	Model card
OpenAI MRCR	52.3%	#6 of 10	Model card
OpenAI MRCR v2	52.3%	#6 of 9	Model card
Ko-LongBench (LG)	86.8%	#2 of 3	Model card
IFBench	67.3%	#49 of 70	Model card
CLIcK	83.9%	#3 of 4	Model card
AA-LCR (Artificial Analysis Long Context Reasoning)	53.5%	#24 of 31	Model card
Humanity's Last Exam	13.6%	#122 of 147	Model card
KMMLU-Pro	67.3%	#6 of 7	Model card

Benchmark	Score	Rank	Source
AIME 2025	92.8%	#35 of 143	Model card
GPQA Diamond	79.1%	#117 of 284	Model card
HRM8K	90.9%	#2 of 3	Model card
HMMT Nov 2025	86.8%	#17 of 23	Model card
IMO-AnswerBench	76.3%	#32 of 37	Model card

Benchmark	Score	Rank	Source
LiveCodeBench v6	80.7%	#27 of 76	Model card
SWE-bench Verified	49.4%	#126 of 150	Model card
CodeUtilityBench (LG)	71.9%	#1 of 1	Model card

Benchmark	Score	Rank	Source
tau2-bench	73.2%	#46 of 69	Model card
BrowseComp	31.4%	#71 of 77	Model card
Terminal-Bench 2.0	29%	#58 of 61	Model card

Benchmark	Score		Rank	Source
KGC-Safety (LG)	96.1%		#2 of 2	Model card
WildJailbreak	89.9%		#2 of 2	Model card

The benchmarks the lab chose to headline, with the scores it gave.

Every result in the longer documents the lab published with it. 2 of them aren't in the launch post.