Grok 4 Is Here: xAI Challenges OpenAI and Google for AI Supremacy
Elon Musk’s xAI team has officially launched Grok 4 , its latest and most advanced multimodal AI model, claiming it now ranks among the top-performing large language models in the industry. According to xAI, Grok 4 delivers state-of-the-art performance across a wide range of academic, mathematical, and reasoning benchmarks — positioning itself as a serious contender against GPT-4o, Gemini 2.5 Pro, and other leading models.
The company also introduced Grok 4 Heavy , an enhanced version that leverages multi-agent tools to achieve even greater results in complex tasks and benchmark tests.
📊 Performance Highlights
xAI shared detailed benchmark results for Grok 4, highlighting its capabilities in both reasoning and technical domains:
🔍 Reasoning & General Intelligence:
- Humanity’s Last Exam (HLE) :
- Grok 4 (no tools): 25.4%
- Grok 4 Heavy (with multi-agent tools): 44.4%
- Gemini 2.5 Pro with tools: 26.9%
- ARC-AGI-2 : Grok 4 scored 16.2% , nearly double the next best model (Claude Opus 4 at ~8.3%).
- MMLU-style evaluations :
- Scored 86.6% overall.
- Achieved an Intelligence Index of 73 , reportedly outperforming OpenAI’s o3 and Google’s Gemini 2.5 Pro.
🧮 STEM & Coding Performance:
- GPQA (Graduate-Level Google Questions):
- Grok 4: 87.5%
- Grok 4 Heavy: 88.9%
- AIME (American Invitational Mathematics Examination):
- Grok 4 Heavy achieved a perfect score of 100%
- Grok 4: 98.8%
- SWE-Bench (Software Engineering Benchmark):
- A specialized variant called Grok 4 Code is set to launch in August 2025 , already achieving scores between 72–75% .
💡 What Makes Grok 4 Special?
According to xAI, Grok 4 is capable of more than just answering questions — it can detect errors or ambiguities in prompts , then either correct them or provide answers for each possible interpretation.
In a recent post on X, Elon Musk emphasized this strength:
“Grok 4 is at the point where it essentially never gets math/physics exam questions wrong, unless they are skillfully adversarial.”
This level of robustness suggests a major leap in reliability and contextual understanding, especially in high-stakes environments like scientific research, engineering, and coding.
💰 Pricing and Subscription Plans
For developers and businesses, Grok 4 API pricing remains unchanged from Grok 3 :
- $3 per million input tokens
- $15 per million output tokens
- $0.75 per million cached input tokens
For individual users, xAI has introduced new subscription tiers:
- Free Tier : Limited access to Grok 3 only.
- SuperGrok ($30/month) : Enhanced access to both Grok 4 and Grok 3.
- SuperGrok Heavy ($300/month) : Full access to Grok 4 Heavy, Grok 4, and Grok 3.
🚀 Final Thoughts
With Grok 4, xAI is clearly aiming to challenge the dominance of established players like OpenAI and Google. Whether it can sustain momentum and gain widespread adoption will depend not only on performance, but also on accessibility, developer support, and real-world use cases.
Still, with these impressive benchmark results and aggressive pricing, Grok 4 looks poised to make a strong impact in the generative AI space.
