Grok-2

xAI

Verified against Artificial Analysis · Jul 13, 2026

Capability

percentile index

Price /M

$0 / $0

in / out

Context

131.1K

tokens

Avg score

55.0

8 benchmarks

AA Intelligence

8.0

index

Weights

Closed

Released

Dec 12, 2024

About

Grok-2 is a proprietary model in the Grok-2 family from xAI. Benchmarked on 8 evals, averaging 55.0.

proprietarytextvision

Benchmark scores · 8

avg 55.0 — higher is better

HumanEvalvendor-reported · 2024-08

88.4

MMLUvendor-reported · 2024-08

87.5

MATHartificial-analysis · 2026-07

77.8

MMLU-Proartificial-analysis · 2026-07

70.9

GPQA Diamondvendor-reported · 2024-08

SciCodeartificial-analysis · 2026-07

28.5

LiveCodeBenchartificial-analysis · 2026-07

26.7

Humanity's Last Examartificial-analysis · 2026-07

3.8

Details

PublisherxAI

FamilyGrok-2

Parameters—

Licenseproprietary

WeightsClosed

ReleasedDec 12, 2024

History

full ledger →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$0/M

xAI131.1K ctx · $10/M