Compare AI models

Specifications
Context window
32k—
Benchmarks
HumanEval
74.4%72%
MATH
53.2%47.9%
MMLU
83.7%81.5%
GSM8K
88.9%87.6%
Overview
CompanyGoogleZ.ai
Release dateFeb 8 2024Jan 16 2024
AccessClosedClosed
Model detailsView modelView model

Frequently asked questions

Gemini 1.0 Ultra leads GLM-4 on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). GLM-4 shipped 23 days before Gemini 1.0 Ultra, so benchmark comparisons should account for the intervening progress.

Published specifications for these two models are limited — see each model page for the latest details.

On HumanEval, Gemini 1.0 Ultra leads at 74.4% vs GLM-4 at 72%. On MATH, Gemini 1.0 Ultra leads at 53.2% vs GLM-4 at 47.9%. On MMLU, Gemini 1.0 Ultra leads at 83.7% vs GLM-4 at 81.5%. On GSM8K, Gemini 1.0 Ultra leads at 88.9% vs GLM-4 at 87.6%.