Compare AI models

Specifications
Parameters
16B14B
Context window
8k2k
Benchmarks
HumanEval
48.1%32.3%
MATH
45.3%24.8%
MMLU
70%66.3%
GSM8K
77.4%61.3%
Overview
CompanyMoonshot AIQwen
Release dateFeb 22 2025Sep 25 2023
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Moonlight leads Qwen-14B on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). Qwen-14B shipped 516 days before Moonlight, so benchmark comparisons should account for the intervening progress.

Moonlight has 16B parameters, while Qwen-14B has 14B. Context windows are 8k (Moonlight) vs 2k (Qwen-14B).

On HumanEval, Moonlight leads at 48.1% vs Qwen-14B at 32.3%. On MATH, Moonlight leads at 45.3% vs Qwen-14B at 24.8%. On MMLU, Moonlight leads at 70% vs Qwen-14B at 66.3%. On GSM8K, Moonlight leads at 77.4% vs Qwen-14B at 61.3%.