EXAONE 4.5 33B — benchmark results

LG AI Research's open 33B vision-language model with hybrid attention (research license). Provider: LG AI. Released 2026-04-09. Access: Open.

Unified ELO 1609 ± 28, rank #461 of 1776 rated models, from 35 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
K-MetBench75.9Accuracy (self-reported)81
AA GPQA Diamond79.39Accuracy (%)72.3
AA IFBench57.96Accuracy (%)69.8
AA TAU-2 Bench78.07Accuracy (%)68.4
AA Humanity's Last Exam11.63Accuracy (%)66.8
Artificial Analysis Intelligence Index22.96Intelligence Index65.1
CritPt0.3Accuracy (self-reported)65
AA-LCR49.3Score (self-reported)63.3
AA Omniscience - Science, Engineering & Mathematics28.8Accuracy (%)60.4
AA Terminal-Bench Hard20.45Accuracy (%)59.9
AA CritPt0.29Accuracy (%)58.7
AA Long Context Reasoning49.33Accuracy (%)57.7

Interactive version: theaggregate.ai/model?slug=exaone-4-5-33b · How the rankings work · Data refreshed daily, snapshot 2026-07-22.