Grok 4, launched on 10 July 2025 by Elon Musk’s company xAI, is being promoted as the most intelligent AI available today.
Marketed with bold claims of PhD-level intelligence across disciplines and the ability to discover new physics within a year, Grok 4 made its debut during a livestream that attracted over 1.5 million viewers on X. The model includes notable upgrades in reasoning, coding, and voice interaction, alongside real-time web browsing through the DeepSearch tool.
Premium access is priced at $300 per month under the SuperGrok Heavy plan.
Grok 4 Reasoning, Coding, and Voice Technology
Grok 4 is built on xAI’s Colossus supercomputer and was trained using 200,000 GPUs. It reportedly scored 25.4 percent on the demanding Humanity’s Last Exam, outperforming Google’s Gemini 2.5 Pro, which achieved 21.6 percent. When given access to external tools, Grok 4 Heavy raised its performance to 44.4 percent.
This strong showing is supported by additional benchmarking on ARC-AGI-2 visual reasoning tasks, where it nearly doubled the accuracy of Claude Opus 4.
In software development, Grok 4 introduces enhanced code-handling abilities.
It supports the debugging and repair of entire source code files, positioning it as a direct competitor to GitHub Copilot, Cursor, and advanced OpenAI models.
For voice interaction, the assistant “Eve” delivers emotionally expressive communication, including British-accented speech, whispering, and even singing. Real-time internet access via DeepSearch further distinguishes Grok 4 in daily use.
Competition and Market Positioning of Grok 4
Grok 4 enters a highly competitive AI landscape alongside GPT-5 from OpenAI, Claude from Anthropic, and Google’s Gemini family. While its 128,000-token context window falls short of Gemini’s 1 million-token capacity, xAI appears to be targeting a niche audience.
This includes developers, researchers, and enterprises looking for high-end reasoning performance over broad consumer polish.
Despite the steep subscription fee, Grok 4 is being noticed for its speed, personality, and capabilities in scientific reasoning.
Critics have pointed out the absence of a formal technical paper or peer-reviewed evaluation, which has been standard for other top-tier AI releases. This raises questions about transparency and repeatability of xAI’s claims.
Future of Grok and xAI’s Vision
Looking ahead, xAI intends to expand Grok’s abilities further. Multimodal functionality, greater alignment with factual truth, and potential integration into robotics are all on the roadmap. The focus is on building an AI system that not only performs well on benchmarks but also interacts in more human and insightful ways.
However, xAI’s challenge remains twofold.
It must deliver on its ambitious performance goals while demonstrating an ability to manage harm, avoid bias, and provide dependable moderation. With Grok 4, Elon Musk’s team has taken a bold step forward, but the path to widespread trust and adoption is far from guaranteed.
Key Features and Claims
- Advanced Reasoning
- Grok 4 reportedly scored 25.4% on the challenging “Humanity’s Last Exam,” surpassing Google’s Gemini 2.5 Pro, which scored 21.6%.
- Musk and xAI claim Grok 4 is “smarter than almost all graduate students, in all disciplines, simultaneously”.
- Coding Capabilities
- Users can paste entire source code files for debugging and fixes, positioning Grok 4 as a direct competitor to tools like GitHub Copilot and Cursor.
- Voice and Real-Time Web Access
- Includes five new voices and real-time web access via xAI’s DeepSearch tool.
Market Context and Comparisons
- Competitive Landscape
- Grok 4’s release comes as OpenAI, Anthropic, and Google are also rolling out advanced AI agents with multi-step reasoning and real-time web browsing.
- Grok 4’s context window is 128,000 tokens, smaller than Gemini’s 1 million.
- User and Industry Response
- Some users praise Grok 4’s capabilities, especially in reasoning and coding, while others criticise its high subscription cost and context window limitations.
Feature Comparison Table
| Feature | Grok 4 (xAI) | Gemini 2.5 Pro (Google) |
|---|---|---|
| Reasoning (Humanity’s Last Exam) | 25.4% | 21.6% |
| Context Window | 128,000 tokens | 1,000,000 tokens |
| Subscription Price | $300/mo (SuperGrok Heavy) | $20–$30/mo (Enterprise) |
| Coding Assistance | Yes | Yes |
| Real-Time Web Access | Yes (DeepSearch) | Yes |
| Voice Options | 5 new voices | Yes |
References
- xAI on X: Grok 4 Launch Announcement (July 2025)
- Cryptobriefing: xAI introduces Grok 4; Elon Musk predicts Grok will start inventing (July 2025)
- Yahoo Finance: Elon Musk’s xAI launches Grok 4 alongside a $300 monthly subscription (July 2025)
- Tom’s Guide: Grok 4 is here—Elon Musk says it’s the same model physicists use (July 2025)
- Teslarati: Elon Musk confirms Grok 4 launch on July 9 with livestream event (July 2025)
- eWeek: Comprehensive Grok Review: Features, Pricing, and Insights (May 2025)
- Dawn: Grok under fire for posts praising Hitler, insulting Islam (July 2025)
- Aloa: Gemini vs Grok – AI Assistant Comparison (2025)
- Moneycontrol: ‘Grok 4 is better than PhDs in every subject,’ Elon Musk claims (July 2025)
- Politico: Linda Yaccarino resigns as CEO of X (July 2025)
- Wired: Elon Musk Unveils Grok 4 Amid Controversy Over Chatbot’s Content (July 2025)