Compare AI models

Specifications
Parameters
7B14.3B
Context window
4k32k
Benchmarks
GSM8K
88.2%61.5%
Overview
CompanyDeepSeekQwen
Release dateFeb 5 2024Mar 28 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek-Math leads Qwen1.5-MoE on 1 of the 1 benchmark they both report (GSM8K). DeepSeek-Math shipped 52 days before Qwen1.5-MoE, so benchmark comparisons should account for the intervening progress.

DeepSeek-Math has 7B parameters, while Qwen1.5-MoE has 14.3B. Context windows are 4k (DeepSeek-Math) vs 32k (Qwen1.5-MoE).

On GSM8K, DeepSeek-Math leads at 88.2% vs Qwen1.5-MoE at 61.5%.