GLM-4.5 Flash: benchmark results

Provider: Zhipu. Access: Open.

Unified ELO 1609 ± 28, rank #412 of 1607 rated models, from 46 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
GABench - Graph ML - Link Prediction (Text-Attributed Graphs)38.09Task success rate (%; share of tasks whose final answer a De80
ReLE - Agents and Tool Use64.1Accuracy (%)71.3
GABench - Graph Retrieval - Edge Level (Text-Attributed Graphs)32.46Task success rate (%; share of tasks whose final answer a De60
GABench - Graph Retrieval - Graph Level (Text-Attributed Graphs)30.63Task success rate (%; share of tasks whose final answer a De60
GABench - Graph Theory - Edge Level (Text-Attributed Graphs)47.92Task success rate (%; share of tasks whose final answer a De60
GABench - Graph Theory - Edge Level (Text-Paired Graphs)16.67Task success rate (%; share of tasks whose final answer a De60
GABench - Graph Theory - Graph Level (Numerical-Attribute Graphs)8.89Task success rate (%; share of tasks whose final answer a De60
GABench - Graph Theory - Graph Level (Text-Attributed Graphs)54.08Task success rate (%; share of tasks whose final answer a De60
GABench - Graph Theory - Graph Level (Text-Paired Graphs)13.57Task success rate (%; share of tasks whose final answer a De60
GABench - Graph Theory - Node Level (Text-Attributed Graphs)53.16Task success rate (%; share of tasks whose final answer a De60
GABench - Open-Ended QA - Prediction (Text-Paired Graphs)6.12Task success rate (%; share of tasks whose final answer a De60
ReLE - Language and Instruction Following65.5Accuracy (%)57.6

Interactive version: theaggregate.ai/model?slug=glm-4-5-flash · How It Works · Data refreshed daily, snapshot 2026-09-29.