Compare AI models

Specifications
Parameters
16B1.5B/7B/72B
Context window
4k4k
Benchmarks
MATH
15.2%84%
GSM8K
62.2%96.7%
Overview
CompanyDeepSeekQwen
Release dateJan 11 2024Aug 8 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Qwen2-Math leads DeepSeek-MoE on 2 of the 2 benchmarks they both report (MATH, GSM8K). DeepSeek-MoE shipped 210 days before Qwen2-Math, so benchmark comparisons should account for the intervening progress.

DeepSeek-MoE has 16B parameters, while Qwen2-Math has 1.5B/7B/72B. Context windows are 4k (DeepSeek-MoE) vs 4k (Qwen2-Math).

On MATH, Qwen2-Math leads at 84% vs DeepSeek-MoE at 15.2%. On GSM8K, Qwen2-Math leads at 96.7% vs DeepSeek-MoE at 62.2%.