Compare AI models

Specifications
Parameters
16B—
Context window
4k1M
Benchmarks
HumanEval
45.7%86.6%
MATH
15.2%81.1%
GSM8K
62.2%93.8%
Overview
CompanyDeepSeekQwen
Release dateJan 11 2024Nov 15 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

Qwen2.5-Turbo (1M) leads DeepSeek-MoE on 3 of the 3 benchmarks they both report (HumanEval, MATH, GSM8K). DeepSeek-MoE shipped 309 days before Qwen2.5-Turbo (1M), so benchmark comparisons should account for the intervening progress.

Context windows are 4k (DeepSeek-MoE) vs 1M (Qwen2.5-Turbo (1M)). DeepSeek-MoE is open weight, while Qwen2.5-Turbo (1M) is closed.

On HumanEval, Qwen2.5-Turbo (1M) leads at 86.6% vs DeepSeek-MoE at 45.7%. On MATH, Qwen2.5-Turbo (1M) leads at 81.1% vs DeepSeek-MoE at 15.2%. On GSM8K, Qwen2.5-Turbo (1M) leads at 93.8% vs DeepSeek-MoE at 62.2%.