Compare AI models

Specifications
Parameters
236B6B
Context window
128k32k
Benchmarks
HumanEval
81.1%9.8%
MATH
53.9%6.9%
MMLU
77.8%45.2%
GSM8K
92.2%25.9%
Overview
CompanyDeepSeekZ.ai
Release dateMay 6 2024Jun 25 2023
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V2 leads ChatGLM2-6B on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). ChatGLM2-6B shipped 316 days before DeepSeek V2, so benchmark comparisons should account for the intervening progress.

DeepSeek V2 has 236B parameters, while ChatGLM2-6B has 6B. Context windows are 128k (DeepSeek V2) vs 32k (ChatGLM2-6B).

On HumanEval, DeepSeek V2 leads at 81.1% vs ChatGLM2-6B at 9.8%. On MATH, DeepSeek V2 leads at 53.9% vs ChatGLM2-6B at 6.9%. On MMLU, DeepSeek V2 leads at 77.8% vs ChatGLM2-6B at 45.2%. On GSM8K, DeepSeek V2 leads at 92.2% vs ChatGLM2-6B at 25.9%.