Compare AI models

Specifications
Parameters
16B14.3B
Context window
4k32k
Benchmarks
HumanEval
45.7%34.2%
MMLU
47.2%62.5%
GSM8K
62.2%61.5%
Overview
CompanyDeepSeekQwen
Release dateJan 11 2024Mar 28 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek-MoE leads Qwen1.5-MoE on 2 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). DeepSeek-MoE shipped 77 days before Qwen1.5-MoE, so benchmark comparisons should account for the intervening progress.

DeepSeek-MoE has 16B parameters, while Qwen1.5-MoE has 14.3B. Context windows are 4k (DeepSeek-MoE) vs 32k (Qwen1.5-MoE).

On HumanEval, DeepSeek-MoE leads at 45.7% vs Qwen1.5-MoE at 34.2%. On MMLU, Qwen1.5-MoE leads at 62.5% vs DeepSeek-MoE at 47.2%. On GSM8K, DeepSeek-MoE leads at 62.2% vs Qwen1.5-MoE at 61.5%.