Compare AI models

Specifications
Parameters
—16B
Context window
1M8k
Benchmarks
HumanEval
74.3%48.1%
MATH
54.9%45.3%
MMLU
78.9%70%
GSM8K
86.2%77.4%
Overview
CompanyGoogleMoonshot AI
Release dateMay 14 2024Feb 22 2025
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

Gemini 1.5 Flash leads Moonlight on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU, GSM8K). Gemini 1.5 Flash shipped 284 days before Moonlight, so benchmark comparisons should account for the intervening progress.

Context windows are 1M (Gemini 1.5 Flash) vs 8k (Moonlight). Gemini 1.5 Flash is closed, while Moonlight is open weight.

On HumanEval, Gemini 1.5 Flash leads at 74.3% vs Moonlight at 48.1%. On MATH, Gemini 1.5 Flash leads at 54.9% vs Moonlight at 45.3%. On MMLU, Gemini 1.5 Flash leads at 78.9% vs Moonlight at 70%. On GSM8K, Gemini 1.5 Flash leads at 86.2% vs Moonlight at 77.4%.