Compare AI models

Specifications
Parameters
16B7B
Context window
4k64k
Benchmarks
HumanEval
45.7%83.5%
Overview
CompanyDeepSeekQwen
Release dateJan 11 2024Apr 16 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

CodeQwen1.5 leads DeepSeek-MoE on 1 of the 1 benchmark they both report (HumanEval). DeepSeek-MoE shipped 96 days before CodeQwen1.5, so benchmark comparisons should account for the intervening progress.

DeepSeek-MoE has 16B parameters, while CodeQwen1.5 has 7B. Context windows are 4k (DeepSeek-MoE) vs 64k (CodeQwen1.5).

On HumanEval, CodeQwen1.5 leads at 83.5% vs DeepSeek-MoE at 45.7%.