Compare AI models

Specifications
Parameters
47B—
Context window
32k128k
Benchmarks
HumanEval
40.2%75.7%
MMLU
70.6%81.9%
GSM8K
58.4%90.9%
Overview
CompanyMistralZ.ai
Release dateDec 8 2023Jun 5 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

GLM-4-Air leads Mixtral 8×7B on 3 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). Mixtral 8×7B shipped 180 days before GLM-4-Air, so benchmark comparisons should account for the intervening progress.

Context windows are 32k (Mixtral 8×7B) vs 128k (GLM-4-Air). Mixtral 8×7B is open weight, while GLM-4-Air is closed.

On HumanEval, GLM-4-Air leads at 75.7% vs Mixtral 8×7B at 40.2%. On MMLU, GLM-4-Air leads at 81.9% vs Mixtral 8×7B at 70.6%. On GSM8K, GLM-4-Air leads at 90.9% vs Mixtral 8×7B at 58.4%.