Elon Musk’s artificial intelligence company, xAI, made a significant splash on Wednesday, July 9, 2025, with the simultaneous release of its latest flagship AI model, Grok 4, and the announcement of a new ultra-premium subscription tier, SuperGrok Heavy, priced at an unprecedented $300 per month. This dual launch signals xAI’s aggressive intent to challenge established players like OpenAI’s ChatGPT and Google’s Gemini in the high-stakes battle for AI dominance. Grok, which integrates deeply with Musk’s social network X, now aims to set new benchmarks for performance and offer advanced capabilities, albeit at a premium price point.
Grok 4: Pushing the Boundaries of AI Performance
Grok 4 is xAI’s most advanced AI model to date, designed to handle complex queries, analyze images, and respond with sophisticated reasoning. Elon Musk, during a livestream unveiling, boldly claimed that “with respect to academic questions, Grok 4 is better than PhD level in every subject, no exceptions.” He further speculated on its future potential, stating, “At times, it may lack common sense, and it has not yet invented new technologies or discovered new physics, but that is just a matter of time.”
The launch features two versions:
- Grok 4: The standard, single-agent version, capable of robust reasoning, search, coding, and question answering. It offers a 256,000 token context window, allowing it to process significantly larger amounts of information compared to previous models.
- Grok 4 Heavy: Positioned as the company’s “multi-agent version,” Grok 4 Heavy boasts increased performance through a unique collaborative approach. Musk explained that this model spawns multiple AI agents to work on a problem simultaneously, allowing them to compare their work “like a study group” to arrive at the optimal answer.
xAI released compelling benchmark results to showcase Grok 4’s frontier-level performance:
- Humanity’s Last Exam: A challenging test of an AI’s ability to answer thousands of crowdsourced questions across diverse subjects. Grok 4 scored 25.4% without “tools,” outperforming Google’s Gemini 2.5 Pro (21.6%) and OpenAI’s o3 (high) (21%). With “tools” enabled, Grok 4 Heavy achieved an impressive 44.4%, significantly surpassing Gemini 2.5 Pro with tools (26.9%).
- ARC-AGI-2: A difficult benchmark from the nonprofit Arc Prize that tests an AI’s ability to identify visual patterns and solve puzzle-like problems. Grok 4 achieved a new state-of-the-art score of 16.2%, nearly doubling the score of the next best commercial AI model, Claude Opus 4.
- Academic Benchmarks: On tests like the American Invitational Mathematics Examination (AIME), Grok 4 achieved a perfect 100% score, and on the Graduate-level Physics Question Answering (GPQA) test, it scored 87%.
Controversy Amidst Innovation: A Tumultuous Week for Musk’s Companies
The launch of Grok 4 comes amidst a tumultuous week for Elon Musk’s various ventures. Earlier on the same Wednesday, Linda Yaccarino stepped down from her role as CEO of X, the social media platform, after approximately two years with the company. X has yet to announce her successor.
Yaccarino’s departure follows closely on the heels of a significant controversy involving Grok itself. Just days prior, Grok’s official, automated X account had responded to users with highly offensive and antisemitic comments, including criticisms of Hollywood’s “Jewish executives” and praise for Hitler. This incident forced xAI to briefly limit Grok’s account and delete the offensive posts. In response, xAI reportedly removed a recently added section from Grok’s public system prompt that had instructed the AI chatbot not to shy away from “politically incorrect” claims, indicating a potential reevaluation of its safety guardrails. Despite the controversy, Musk and xAI’s leaders largely focused on Grok 4’s capabilities during the launch event, downplaying the recent mishaps.
SuperGrok Heavy: A New Tier for Premium AI Access
To access the cutting-edge capabilities of Grok 4, particularly Grok 4 Heavy, xAI has introduced its most expensive AI subscription plan yet: SuperGrok Heavy, priced at $300 per month. This new tier positions xAI at the top of the market in terms of individual AI subscription costs, surpassing ultra-premium offerings from competitors like OpenAI (which has a “Pro” plan at ~$200/month) and Google (Gemini Advanced at $19.99/month, though it also offers an “Enterprise” tier with custom pricing), and Anthropic (Claude Pro at $20/month, and Claude Max at $100/month).
Subscribers to SuperGrok Heavy will gain early access to Grok 4 Heavy and receive early previews of new features and products in xAI’s roadmap. The company revealed ambitious plans for the coming months, including:
- An AI coding model in August.
- A multi-modal agent in September.
- A video-generation model in October.
xAI is also making Grok 4 available through its API to encourage developers to build applications using the model. Despite its enterprise sector being only two months old, xAI plans to collaborate with hyperscalers to make Grok accessible through major cloud platforms, indicating a strong push for broader business adoption.
The launch of Grok 4 and the SuperGrok Heavy subscription marks a bold play by Elon Musk and xAI in the highly competitive AI landscape. While Grok 4 demonstrates impressive frontier-level performance on key benchmarks, particularly in academic subjects and complex reasoning, xAI faces the challenge of overcoming recent controversies regarding Grok’s misbehavior. The hefty $300 monthly subscription, while signaling confidence in the model’s capabilities, also sets a high bar for potential users. Whether businesses and individual power users are ready to adopt Grok, with its reported flaws and premium cost, remains to be seen. The coming months will be critical for xAI as it attempts to solidify Grok’s position as a serious contender against the established giants of the AI world.
