Compare AI models

Specifications
Parameters
16B/236B16B
Context window
128k8k
Benchmarks
LiveCodeBench
43.4%14.3%
HumanEval
90.2%48.1%
MATH
75.7%45.3%
MMLU
79.2%70%
GSM8K
94.9%77.4%
Overview
CompanyDeepSeekMoonshot AI
Release dateJun 14 2024Feb 22 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek Coder V2 leads Moonlight on 5 of the 5 benchmarks they both report. DeepSeek Coder V2 shipped 253 days before Moonlight, so benchmark comparisons should account for the intervening progress.

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

On LiveCodeBench, DeepSeek Coder V2 leads at 43.4% vs Moonlight at 14.3%. On HumanEval, DeepSeek Coder V2 leads at 90.2% vs Moonlight at 48.1%. On MATH, DeepSeek Coder V2 leads at 75.7% vs Moonlight at 45.3%. On MMLU, DeepSeek Coder V2 leads at 79.2% vs Moonlight at 70%. On GSM8K, DeepSeek Coder V2 leads at 94.9% vs Moonlight at 77.4%.