Compare AI models

Specifications
Parameters
—16B
Context window
128k8k
API pricing
Cheapest input
$0.40Mistral—
Cheapest output
$2.00Mistral—
$180Mistral · zdr—
Benchmarks
LiveCodeBench
30.3%14.3%
HumanEval
92.1%48.1%
GPQA Diamond
57.1%35.2%
MMLU-Pro
77.2%42.4%
Overview
CompanyMistralMoonshot AI
Release dateMay 7 2025Feb 22 2025
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

Mistral Medium 3 leads Moonlight on 4 of the 4 benchmarks they both report (LiveCodeBench, HumanEval, GPQA Diamond, MMLU-Pro). Moonlight shipped 74 days before Mistral Medium 3, so benchmark comparisons should account for the intervening progress.

Context windows are 128k (Mistral Medium 3) vs 8k (Moonlight). Mistral Medium 3 is closed, while Moonlight is open weight.

On LiveCodeBench, Mistral Medium 3 leads at 30.3% vs Moonlight at 14.3%. On HumanEval, Mistral Medium 3 leads at 92.1% vs Moonlight at 48.1%. On GPQA Diamond, Mistral Medium 3 leads at 57.1% vs Moonlight at 35.2%. On MMLU-Pro, Mistral Medium 3 leads at 77.2% vs Moonlight at 42.4%.