GLM-5.2 (Non-reasoning): benchmark results

GLM-5.2 evaluated with reasoning disabled. Provider: Zhipu. Released 2026-06-13. Access: Open.

Unified ELO 1606 ± 1, rank #408 of 1761 rated models, from 25 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
UGI - Writing61.21Writing Score93.5
UGI - Natural Intelligence54.88NatInt Score91.9
AA Omniscience-6.57Score76.4
Artificial Analysis Intelligence Index26.45Intelligence Index75.2
AA CritPt3.14Accuracy (%)74.6
AA GDPval1311.7ELO72.2
FrontierMath - Tiers 1-3 (v2)42.46Accuracy (%, 285 private v2 problems)62.2
AA Omniscience - Software Engineering (SWE)31.5Accuracy (%)60
AA Omniscience - Law13.4Accuracy (%)57.2
AA Humanity's Last Exam9.78Accuracy (%)53.2
Epoch AI - Mystery Game Puzzles18Score52.8
AA Omniscience - Humanities & Social Sciences20.6Accuracy (%)51.7

Interactive version: theaggregate.ai/model?slug=glm-5-2-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.