Compare AI models

Specifications
Parameters
7B/13B/70B16B
Context window
4k8k
Benchmarks
HumanEval
29.9%48.1%
MATH
13.5%45.3%
WinoGrande
80.2%74.6%
MMLU
68.9%70%
GSM8K
56.8%77.4%
Overview
CompanyMetaMoonshot AI
Release dateJul 18 2023Feb 22 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Moonlight leads LLaMA 2 on 4 of the 5 benchmarks they both report. LLaMA 2 shipped 585 days before Moonlight, so benchmark comparisons should account for the intervening progress.

LLaMA 2 has 7B/13B/70B parameters, while Moonlight has 16B. Context windows are 4k (LLaMA 2) vs 8k (Moonlight).

On HumanEval, Moonlight leads at 48.1% vs LLaMA 2 at 29.9%. On MATH, Moonlight leads at 45.3% vs LLaMA 2 at 13.5%. On WinoGrande, LLaMA 2 leads at 80.2% vs Moonlight at 74.6%. On MMLU, Moonlight leads at 70% vs LLaMA 2 at 68.9%. On GSM8K, Moonlight leads at 77.4% vs LLaMA 2 at 56.8%.