This model is 18 months old. A newer version of it has since shipped: Grok 4.6 (2026-08-12). The details below are still accurate for this model — it just is not what we would recommend today.

Grok 3

xAI🇺🇸 United States
active

xAI flagship with real-time X/Twitter data access and strong reasoning.

Context window131K tokens
Input / 1M tokens$3
Output / 1M tokens$15

Version History

grok-3-2025-10minor

Grok 3 October update extending knowledge cutoff and improving accuracy on real-time query grounding from X data.

grok-3-betamajor

Grok 3 beta. xAI claims #1 on AIME, GPQA, and LiveCodeBench at launch.

Benchmark Scores

Full leaderboard →
83.9%
AIME 2024
82.7%
AIME 2025
75.3%
GPQA
5.8%
Hallucination (Summarization)
91.2%
HumanEval
58.4%
LiveCodeBench
97.6%
MATH
87.5%
MMLU
78.1%
MMLU-Pro
55.0 tokens_per_sec
Speed (tok/s)
49.5%
SWE-bench Verified