Atlas

Models

← All models

Grok 4

xAI · 2025-07-09

Grok 4 is xAI's flagship model released in July 2025, trained on xAI's 200,000-GPU Colossus cluster with reasoning reinforcement learning run at pretraining scale. It features native tool use, real-time web search integration, and a 256,000-token context window, and was released alongside Grok 4 Heavy, a multi-agent configuration that coordinates multiple Grok 4 instances for deeper analysis. The model is available via the xAI API and SuperGrok subscriptions.

Benchmark scores

BenchmarkScore
AA-LCR68.0
AA-LCR v1.168.0
AA-Omniscience Index2.1
Active Probing Benchmark (APB)17.6
Active Probing Benchmark (Biology)14.2
Active Probing Benchmark (Chemistry)16.7
Active Probing Benchmark (Computer Science)13.3
Active Probing Benchmark (Economics)20.8
Active Probing Benchmark (Human Disease)30.8
Active Probing Benchmark (Mathematics)10.0
Active Probing Benchmark (Philosophy)21.7
Active Probing Benchmark (Physics)13.3
Agent Red Teaming k=100 (CAIS)85.9
Aider Polyglot79.6
AIME 2024 (avg@8)93.3
AIME 2024-202590.6
AIME 202591.7
AIME 2025 (Artificial Analysis avg@10)92.7
AIME 2025 (avg@8)87.8
AIME 2025 (MathArena repeated-run mean)92.5
Apex (MathArena)2.1
APEX (Mercor)8.2
Apex Shortlist57.5
APEX-SWE21.0
APEX-v1-extended57.1
ARC-AGI-1 Semi-Private66.7
ARC-AGI-216.0
Arena Text — Business, Management & Financial Ops (No Style Control)1388
Arena Text — Chinese (No Style Control)1427
Arena Text — Creative Writing (No Style Control)1397
Loading Atlas data…