Compare AI models

Specifications
Parameters
6B—
Context window
8k—
Benchmarks
HumanEval
58.5%72%
MATH
25.7%47.9%
MMLU
61.4%81.5%
GSM8K
72.3%87.6%
Overview
CompanyZ.aiZ.ai
Release dateOct 27 2023Jan 16 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

GLM-4 leads ChatGLM3-6B on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). ChatGLM3-6B shipped 81 days before GLM-4, so benchmark comparisons should account for the intervening progress.

ChatGLM3-6B is open weight, while GLM-4 is closed.

On HumanEval, GLM-4 leads at 72% vs ChatGLM3-6B at 58.5%. On MATH, GLM-4 leads at 47.9% vs ChatGLM3-6B at 25.7%. On MMLU, GLM-4 leads at 81.5% vs ChatGLM3-6B at 61.4%. On GSM8K, GLM-4 leads at 87.6% vs ChatGLM3-6B at 72.3%.