Grok 4
Parameters
1.7T
Context Length
256K
Modality
Multimodal
Architecture
Dense
License
Proprietary
Release Date
9 Jul 2025
Knowledge Cutoff
Dec 2024
Evaluation Benchmarks
Rank
#9
| Benchmark | Score | Rank |
|---|---|---|
QA Assistant | 0.985 | 🥇 1 |
Summarization | 0.976 | 🥉 3 |
StackUnseen | 0.886 | 5 |
Coding | 0.80 | 6 |
Graduate-Level QA | 0.875 | 6 |
Professional Knowledge | 0.85 | 19 |
Rankings
Overall Rank
#9
Coding Rank
#3 🥉
About Grok 4
Grok 4 is xAI's most intelligent model, trained with reinforcement learning at unprecedented scale using the 200,000 GPU Colossus cluster. Features native tool use with code interpreter and web browsing, real-time search integration across X and the web, and state-of-the-art performance on Humanity's Last Exam (50.7% text-only subset with tools). Achieves 100% on AIME 2025, 99.4% on HMMT 2025, 15.9% on ARC-AGI-2, and dominates Vending-Bench with $4694.15 net worth. Supports 256K context window with multimodal understanding. Available through API with advanced reasoning, coding, and visual processing capabilities.
Technical Specifications
Detailed architecture specifications are limited for proprietary models.
Resources
About Grok 4
xAI's frontier intelligence models trained with reinforcement learning at unprecedented scale using the 200,000 GPU Colossus cluster. Grok 4 series demonstrates state-of-the-art performance in reasoning, coding, and multimodal understanding with native tool use capabilities. Features real-time search integration across X and the web, advanced reasoning through scaled RL training, and industry-leading performance on academic benchmarks. Designed for both immediate responses and extended thinking modes with vision capabilities.