On July 10, 2025, Elon Musk and his AI company xAI unveiled Grok 4, which Musk described as “the smartest artificial intelligence in the world.” The model, particularly in its premium Grok 4 Heavy version, aims to outperform existing large language models with superior reasoning, multi-agent architecture, and bold claims of future innovation.
Grok 4 is the latest iteration of the xAI chatbot, originally designed to compete with OpenAI's ChatGPT and Google Gemini. According to Musk and the xAI team, Grok 4 has surpassed the PhD level in all disciplines and may soon begin to “discover new technologies” before the end of 2025.
The launch event also introduced a premium subscription model, priced at $300/month (around €255), offering access to Grok 4 Heavy, the company’s most powerful AI system.
Musk claimed Grok 4 performs at or above the PhD level in various academic domains. The system reportedly handles complex questions across mathematics, science, engineering, and linguistics—sometimes outperforming human experts.
The most notable technical advancement in Grok 4 Heavy is its multi-agent system, which enables different AI agents to tackle tasks simultaneously and cross-check responses. Musk described this as creating a “virtual study group,” improving both accuracy and reliability.
xAI showcased Grok 4’s results on several academic and cognitive benchmarks:
These results place Grok 4 at the top of the current generative AI leaderboard.
While Grok 4 currently lacks advanced image generation capabilities—Musk said the system is still “partially blind”—xAI plans to introduce multimodal capabilities, code generation models, and even video-generation AI by late 2025.
xAI has launched Grok 4 with a two-tier pricing strategy:
This positions xAI’s model as the most expensive mainstream AI subscription, targeting:
The pricing strategy contrasts with competitors like OpenAI’s GPT-4o or Anthropic’s Claude, which offer access at significantly lower rates.
Recent controversies marred the Grok launch. Earlier versions of the chatbot generated offensive content, including:
These incidents sparked public backlash and raised serious concerns about the lack of robust content moderation in xAI products.
Just one day before the Grok 4 reveal, Linda Yaccarino, the CEO of X (formerly Twitter), resigned amid ongoing platform controversies—many related to Grok’s content moderation. This has led to broader questions about leadership stability and ethical oversight at xAI and X Corp.
| Feature | Grok 4 Heavy | GPT-4o (OpenAI) | Gemini 2.5 (Google) | Claude Opus (Anthropic) |
| Multi-Agent System | Yes | No | No | No |
| Price | $300/month | $20/month | ~$20/month | ~$20/month |
| ARC-AGI-2 Score | 15.9–16.2% | ~10% | ~7–8% | ~8–9% |
| Tool Use Benchmark | 44.4% | Not disclosed | 26.9% | ~35% (est.) |
| Modality | Text (Multimodal coming soon) | Fully multimodal | Multimodal | Fully multimodal |
Grok 4 leads in reasoning benchmarks but lags in multimodal performance—a major growth area for 2025.
Despite its high scores and strong architectural design, Grok 4 is not yet capable of independent technological discovery or scientific theory formulation. However, Musk remains confident:
“I’d be surprised if it hasn’t discovered something new by next year.”
—Elon Musk
If this happens, Grok 4 would mark a true inflection point in the development of creative AI, crossing from generative to innovative intelligence.
Grok 4 represents xAI’s most ambitious effort yet:
Whether Grok 4 becomes a revolutionary innovation tool or merely a premium product for niche users depends on how xAI addresses its current limitations—both technical and ethical.
By continuing to use the site, you agree to the use of cookies. more information
The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.
