GLM-4 9B (0414): benchmark results

Provider: Zhipu. Released 2024-06-05. Access: Open.

Unified ELO 1479 ± 45, rank #761 of 1537 rated models, from 16 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open Portuguese LLM - FaQuAD NLI86.86Macro F1 (%)99.6
Open Portuguese LLM - ASSIN2 RTE93.79Macro F1 (%)92
Open Portuguese LLM - ENEM74.32Accuracy (%)86.7
Open Portuguese LLM - BLUEX63.14Accuracy (%)83.1
Open Portuguese LLM - OAB Exams49.66Accuracy (%)73.9
CommunityBench - Community-Consistent Generation176.85BTL-Elo rating from majority-voted LLM-judge pairwise compar56.2
CommunityBench - Community Identification51.25Accuracy (%)37.5
CommunityBench - Preference Distribution Kendall Tau0.06Kendall's Tau (-1 to 1)31.2
CommunityBench - Preference Identification33.12Accuracy (%)25
CommunityBench - Preference Distribution JSD0.17Jensen-Shannon Divergence (0-1)18.8
ReLE - Education - Primary School Subjects30Accuracy (%)7.1
ReLE - Education - High School Subjects27.1Accuracy (%)6

Interactive version: theaggregate.ai/model?slug=glm-4-9b-0414 · How It Works · Data refreshed daily, snapshot 2026-09-25.