Compare AI models

Specifications
Parameters
7B/67B6B
Context window
4k32k
Benchmarks
HumanEval
73.8%9.8%
MATH
32.6%6.9%
MMLU
71.1%45.2%
GSM8K
84.1%25.9%
Overview
CompanyDeepSeekZ.ai
Release dateNov 29 2023Jun 25 2023
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek-LLM leads ChatGLM2-6B on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). ChatGLM2-6B shipped 157 days before DeepSeek-LLM, so benchmark comparisons should account for the intervening progress.

DeepSeek-LLM has 7B/67B parameters, while ChatGLM2-6B has 6B. Context windows are 4k (DeepSeek-LLM) vs 32k (ChatGLM2-6B).

On HumanEval, DeepSeek-LLM leads at 73.8% vs ChatGLM2-6B at 9.8%. On MATH, DeepSeek-LLM leads at 32.6% vs ChatGLM2-6B at 6.9%. On MMLU, DeepSeek-LLM leads at 71.1% vs ChatGLM2-6B at 45.2%. On GSM8K, DeepSeek-LLM leads at 84.1% vs ChatGLM2-6B at 25.9%.