Compare AI models

Specifications
Parameters
24B16B
Context window
128k8k
API pricing
Cheapest input
$0.06DeepInfra—
Cheapest output
$0.16DeepInfra—
$25.00DeepInfra · fp8—
Benchmarks
MATH
69.42%45.3%
GPQA Diamond
46.13%35.2%
SimpleQA
12.1%5.6%
MMLU-Pro
69.06%42.4%
MMLU
80.5%70%
Overview
CompanyMistralMoonshot AI
Release dateJun 20 2025Feb 22 2025
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Mistral Small 3.2 leads Moonlight on 5 of the 5 benchmarks they both report. Moonlight shipped 118 days before Mistral Small 3.2, so benchmark comparisons should account for the intervening progress.

Mistral Small 3.2 has 24B parameters, while Moonlight has 16B. Context windows are 128k (Mistral Small 3.2) vs 8k (Moonlight).

On MATH, Mistral Small 3.2 leads at 69.42% vs Moonlight at 45.3%. On GPQA Diamond, Mistral Small 3.2 leads at 46.13% vs Moonlight at 35.2%. On SimpleQA, Mistral Small 3.2 leads at 12.1% vs Moonlight at 5.6%. On MMLU-Pro, Mistral Small 3.2 leads at 69.06% vs Moonlight at 42.4%. On MMLU, Mistral Small 3.2 leads at 80.5% vs Moonlight at 70%.