Compare AI models

Specifications
Parameters
16B—
Context window
4k128k
Benchmarks
HumanEval
45.7%75.7%
MATH
15.2%57.9%
MMLU
47.2%81.9%
GSM8K
62.2%90.9%
Overview
CompanyDeepSeekZ.ai
Release dateJan 11 2024Jun 5 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

GLM-4-Air leads DeepSeek-MoE on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). DeepSeek-MoE shipped 146 days before GLM-4-Air, so benchmark comparisons should account for the intervening progress.

Context windows are 4k (DeepSeek-MoE) vs 128k (GLM-4-Air). DeepSeek-MoE is open weight, while GLM-4-Air is closed.

On HumanEval, GLM-4-Air leads at 75.7% vs DeepSeek-MoE at 45.7%. On MATH, GLM-4-Air leads at 57.9% vs DeepSeek-MoE at 15.2%. On MMLU, GLM-4-Air leads at 81.9% vs DeepSeek-MoE at 47.2%. On GSM8K, GLM-4-Air leads at 90.9% vs DeepSeek-MoE at 62.2%.