EXAONE 4.5 33B

EXAONE 4.5 33B places in the top three on 2 benchmarks.

LG AI Research Released 9 Apr 2026 Family 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 Apr 2026

Open weights Mid-size

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
IFEval	89.6%	#31 of 126	Model card
MMLU-Pro	83.3%	#47 of 151	Model card
MMMLU	85.4%	#31 of 62	Model card
HallusionBench	63.7%	#14 of 23	Model card
KoBALT	52.1%	#2 of 3	Model card
KMMLU-Pro	67.6%	#5 of 7	Model card
IFBench	62.6%	#52 of 70	Model card
AA-LCR (Artificial Analysis Long Context Reasoning)	50.6%	#27 of 31	Model card

Benchmark	Score	Rank	Source
AIME 2025	92.9%	#34 of 143	Model card
GPQA Diamond	80.5%	#109 of 284	Model card
AIME 2026	92.6%	#22 of 34	Model card

Benchmark	Score		Rank	Source
LiveCodeBench v6	81.4%		#26 of 76	Model card

Benchmark	Score		Rank	Source
tau2-bench	72%		#51 of 69	Model card

Benchmark	Score		Rank	Source
MedXpertQA-MM	42.1%		#11 of 12	Model card

Benchmark	Score	Rank	Source
MathVista	85%	#14 of 64	Model card
MMMU	78.7%	#22 of 97	Model card
OmniDocBench v1.5	81.2	#3 of 13	Model card
BLINK	68.8%	#7 of 28	Model card
We-Math	79.1%	#3 of 11	Model card
LogicVista	73.8%	#4 of 12	Model card
OCRBench v2	63.2%	#8 of 20	Model card
AI2D	89%	#19 of 44	Model card
MATH-Vision (MathVision)	75.2%	#19 of 42	Model card
MMStar	74.9%	#18 of 30	Model card
MMMU-Pro	68.6%	#45 of 72	Model card
CharXiv	71.7%	#49 of 67	Model card
ChartQA Pro	62.2%	#7 of 8	Model card
BabyVision	18.8%	#20 of 20	Model card
K-Viscuit	80.1%	#1 of 1	Model card
KMMMU	42.7%	#1 of 1	Model card
KRETA	91.9%	#1 of 1	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.