Zhipu (GLM) · model
GLM-5.3 (max) benchmark scores
As of 2026-10-07, GLM-5.3 (max) (Zhipu (GLM)) has measured scores on 12 benchmarks and estimated scores on 3 more.
| Benchmark | Score | Source |
|---|---|---|
| Terminal-Bench 2.1 | 83.9% | measured |
| Terminal-Bench 4.0 | 41.9% | measured |
| Terminal-Bench Science 0.1 | 9.5% | measured |
| GPQA Diamond | 91.7% | measured |
| SciCode | 59.0% | measured |
| CritPt | 19.1% | measured |
| GDP.pdf | 11.2% | measured |
| τ³-Bench Banking | 50.3% | measured |
| MMMU-Pro | 75.3% | estimated ± 2.6 pp, high confidence |
| AA-AnalystAgent | 19.4% | estimated ± 3.6 pp, medium confidence |
| Harvey LAB | 91.2% | estimated ± 1.2 pp, medium confidence |
| Humanity's Last Exam | 42.3% | measured |
| AutomationBench | 62.2% | measured |
| EnterpriseOps-Gym | 36.4% | measured |
| ITBench SRE | 46.1% | measured |