Compare AI models

Specifications
Parameters
16B—
Context window
8k128k
Benchmarks
HumanEval
48.1%88.4%
MATH
45.3%76.1%
MMLU-Pro
42.4%75.5%
MMLU
70%87.5%
Overview
CompanyMoonshot AISpaceXSI
Release dateFeb 22 2025Aug 13 2024
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

Grok‑2 leads Moonlight on 4 of the 4 benchmarks they both report (HumanEval, MATH, MMLU-Pro, MMLU). Grok‑2 shipped 193 days before Moonlight, so benchmark comparisons should account for the intervening progress.

Context windows are 8k (Moonlight) vs 128k (Grok‑2). Moonlight is open weight, while Grok‑2 is closed.

On HumanEval, Grok‑2 leads at 88.4% vs Moonlight at 48.1%. On MATH, Grok‑2 leads at 76.1% vs Moonlight at 45.3%. On MMLU-Pro, Grok‑2 leads at 75.5% vs Moonlight at 42.4%. On MMLU, Grok‑2 leads at 87.5% vs Moonlight at 70%.