Compare AI models

Specifications
Parameters
7B14B
Context window
4k2k
Benchmarks
MATH
51.7%24.8%
GSM8K
88.2%61.3%
Overview
CompanyDeepSeekQwen
Release dateFeb 5 2024Sep 25 2023
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek-Math leads Qwen-14B on 2 of the 2 benchmarks they both report (MATH, GSM8K). Qwen-14B shipped 133 days before DeepSeek-Math, so benchmark comparisons should account for the intervening progress.

DeepSeek-Math has 7B parameters, while Qwen-14B has 14B. Context windows are 4k (DeepSeek-Math) vs 2k (Qwen-14B).

On MATH, DeepSeek-Math leads at 51.7% vs Qwen-14B at 24.8%. On GSM8K, DeepSeek-Math leads at 88.2% vs Qwen-14B at 61.3%.