Compare AI models

Specifications
Parameters
72B—
Context window
32k—
Benchmarks
HumanEval
35.4%72%
MATH
35.2%47.9%
MMLU
77.4%81.5%
GSM8K
78.9%87.6%
Overview
CompanyQwenZ.ai
Release dateNov 30 2023Jan 16 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

GLM-4 leads Qwen-72B on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). Qwen-72B shipped 47 days before GLM-4, so benchmark comparisons should account for the intervening progress.

Qwen-72B is open weight, while GLM-4 is closed.

On HumanEval, GLM-4 leads at 72% vs Qwen-72B at 35.4%. On MATH, GLM-4 leads at 47.9% vs Qwen-72B at 35.2%. On MMLU, GLM-4 leads at 81.5% vs Qwen-72B at 77.4%. On GSM8K, GLM-4 leads at 87.6% vs Qwen-72B at 78.9%.