Compare AI models

Specifications
Parameters
7B/67B16B
Context window
4k8k
Benchmarks
HumanEval
73.8%48.1%
MATH
32.6%45.3%
MMLU
71.1%70%
GSM8K
84.1%77.4%
Overview
CompanyDeepSeekMoonshot AI
Release dateNov 29 2023Feb 22 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek-LLM leads Moonlight on 3 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). DeepSeek-LLM shipped 451 days before Moonlight, so benchmark comparisons should account for the intervening progress.

DeepSeek-LLM has 7B/67B parameters, while Moonlight has 16B. Context windows are 4k (DeepSeek-LLM) vs 8k (Moonlight).

On HumanEval, DeepSeek-LLM leads at 73.8% vs Moonlight at 48.1%. On MATH, Moonlight leads at 45.3% vs DeepSeek-LLM at 32.6%. On MMLU, DeepSeek-LLM leads at 71.1% vs Moonlight at 70%. On GSM8K, DeepSeek-LLM leads at 84.1% vs Moonlight at 77.4%.