Compare AI models

Specifications
Parameters
16B7B/14B
Context window
8k1M
Benchmarks
LiveCodeBench
14.3%38.6%
HumanEval
48.1%88.4%
MATH
45.3%79.5%
MMLU-Pro
42.4%63.3%
GSM8K
77.4%94.8%
Overview
CompanyMoonshot AIQwen
Release dateFeb 22 2025Jan 27 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Qwen2.5-1M leads Moonlight on 5 of the 5 benchmarks they both report. Qwen2.5-1M shipped 26 days before Moonlight, so benchmark comparisons should account for the intervening progress.

Moonlight has 16B parameters, while Qwen2.5-1M has 7B/14B. Context windows are 8k (Moonlight) vs 1M (Qwen2.5-1M).

On LiveCodeBench, Qwen2.5-1M leads at 38.6% vs Moonlight at 14.3%. On HumanEval, Qwen2.5-1M leads at 88.4% vs Moonlight at 48.1%. On MATH, Qwen2.5-1M leads at 79.5% vs Moonlight at 45.3%. On MMLU-Pro, Qwen2.5-1M leads at 63.3% vs Moonlight at 42.4%. On GSM8K, Qwen2.5-1M leads at 94.8% vs Moonlight at 77.4%.