Compare AI models

Specifications
Parameters
16B7B/14B
Context window
32k1M
Benchmarks
HumanEval
57.3%88.4%
MATH
27.9%79.5%
GSM8K
72%94.8%
Overview
CompanyDeepSeekQwen
Release dateMay 16 2024Jan 27 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Qwen2.5-1M leads DeepSeek V2 Lite on 3 of the 3 benchmarks they both report (HumanEval, MATH, GSM8K). DeepSeek V2 Lite shipped 256 days before Qwen2.5-1M, so benchmark comparisons should account for the intervening progress.

DeepSeek V2 Lite has 16B parameters, while Qwen2.5-1M has 7B/14B. Context windows are 32k (DeepSeek V2 Lite) vs 1M (Qwen2.5-1M).

On HumanEval, Qwen2.5-1M leads at 88.4% vs DeepSeek V2 Lite at 57.3%. On MATH, Qwen2.5-1M leads at 79.5% vs DeepSeek V2 Lite at 27.9%. On GSM8K, Qwen2.5-1M leads at 94.8% vs DeepSeek V2 Lite at 72%.