Compare AI models

Specifications
Parameters
70B16B
Context window
128k8k
API pricing
Cheapest input
$0.10DeepInfra—
Cheapest output
$0.32DeepInfra—
$36.00DeepInfra · turbo · fp8—
Benchmarks
HumanEval
88.4%48.1%
MATH
77%45.3%
GPQA Diamond
50.5%35.2%
MMLU-Pro
68.9%42.4%
MMLU
86%70%
Overview
CompanyMetaMoonshot AI
Release dateDec 6 2024Feb 22 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

LLaMA 3.3 leads Moonlight on 5 of the 5 benchmarks they both report. LLaMA 3.3 shipped 78 days before Moonlight, so benchmark comparisons should account for the intervening progress.

LLaMA 3.3 has 70B parameters, while Moonlight has 16B. Context windows are 128k (LLaMA 3.3) vs 8k (Moonlight).

On HumanEval, LLaMA 3.3 leads at 88.4% vs Moonlight at 48.1%. On MATH, LLaMA 3.3 leads at 77% vs Moonlight at 45.3%. On GPQA Diamond, LLaMA 3.3 leads at 50.5% vs Moonlight at 35.2%. On MMLU-Pro, LLaMA 3.3 leads at 68.9% vs Moonlight at 42.4%. On MMLU, LLaMA 3.3 leads at 86% vs Moonlight at 70%.