Elon Musk’s xAI has officially launched Grok 4, its most advanced AI model to date, positioning it as a direct competitor to OpenAI’s GPT-5 and Google’s Gemini 2.5 Pro. Unveiled during a livestream on X (formerly Twitter), Grok 4 is described by Musk as “smarter than almost every graduate student across all fields,” and is designed to push the boundaries of AI reasoning, scientific problem-solving, and real-time information access.
Grok 4 leverages the immense computing power of xAI’s Colossus supercomputer, having been trained on over 200,000 GPUs—ten times the scale of its predecessor, Grok 3. This has resulted in a dramatic leap in reasoning power and benchmark performance, with Grok 4 achieving top scores on industry tests like ARC-AGI and Humanity’s Last Exam, and outperforming leading competitors in academic, mathematical, and coding benchmarks.
Key Features and Capabilities
- Two Variants:
- Multimodal Support:
- Heavy Version:
- Developer Tools:
- Security and API:
- Use Cases:
Musk claims Grok 4 is at a “postgraduate, PhD level in every subject—better than PhD, no exceptions,” and suggests the AI could begin discovering new technologies by late 2025 and potentially uncover new physics within two years.
Pricing and Availability
Grok 4 is available through two subscription tiers:
- SuperGrok Heavy: $300/month — includes access to Grok 4 Heavy (multi-agent), higher rate limits, and early feature access
- SuperGrok: $30/month — includes general access to Grok 4 and Grok 3
The model is also accessible via xAI’s API and is being deployed across hyperscalers and enterprise environments, with further multimodal and video generation tools planned for release later in 2025.
With Grok 4, xAI is staking a claim at the forefront of advanced, multimodal artificial intelligence, aiming to redefine what’s possible in both generalist and developer-focused AI applications.