Compare AI models

Specifications
Parameters
236B16B
Context window
128k8k
Benchmarks
HumanEval
84.76%48.1%
MATH
71.02%45.3%
Overview
CompanyDeepSeekMoonshot AI
Release dateJun 28 2024Feb 22 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V2-0628 leads Moonlight on 2 of the 2 benchmarks they both report (HumanEval, MATH). DeepSeek V2-0628 shipped 239 days before Moonlight, so benchmark comparisons should account for the intervening progress.

DeepSeek V2-0628 has 236B parameters, while Moonlight has 16B. Context windows are 128k (DeepSeek V2-0628) vs 8k (Moonlight).

On HumanEval, DeepSeek V2-0628 leads at 84.76% vs Moonlight at 48.1%. On MATH, DeepSeek V2-0628 leads at 71.02% vs Moonlight at 45.3%.