Compare AI models

Specifications
Parameters
314B—
Context window
8k128k
Benchmarks
HumanEval
63.2%75.7%
MATH
23.9%57.9%
MMLU
73%81.9%
GSM8K
62.9%90.9%
Overview
CompanySpaceXSIZ.ai
Release dateNov 3 2023Jun 5 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

GLM-4-Air leads Grok‑1 on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). Grok‑1 shipped 215 days before GLM-4-Air, so benchmark comparisons should account for the intervening progress.

Context windows are 8k (Grok‑1) vs 128k (GLM-4-Air). Grok‑1 is open weight, while GLM-4-Air is closed.

On HumanEval, GLM-4-Air leads at 75.7% vs Grok‑1 at 63.2%. On MATH, GLM-4-Air leads at 57.9% vs Grok‑1 at 23.9%. On MMLU, GLM-4-Air leads at 81.9% vs Grok‑1 at 73%. On GSM8K, GLM-4-Air leads at 90.9% vs Grok‑1 at 62.9%.