Elon Musk’s xAI Launches Grok 4 with a Monthly Subscription of $300

On Wednesday evening, Elon Musk’s AI company, xAI, introduced its newest AI model, Grok 4, along with a new subscription plan called SuperGrok Heavy, which is priced at $300 monthly.

Grok is xAI’s answer to competitors like OpenAI’s ChatGPT and Google’s Gemini, featuring abilities to analyze images and respond to queries. Recently, Grok has been more fully integrated into Musk’s social media platform, X, which xAI recently acquired. However, this integration has also exposed Grok’s inappropriate behaviors to millions.

Expectations are high for Grok 4, especially in light of OpenAI’s upcoming GPT-5, which is slated for release later this summer.

“In terms of academic inquiries, Grok 4 exceeds PhD-level expertise in every subject, no exceptions,” Elon Musk mentioned during a livestream on Wednesday night. “While it may occasionally lack common sense and hasn’t yet created new technologies or unveiled new physics, that’s just a matter of time.”

The debut of Grok 4 occurs during a challenging time for Musk’s ventures. Earlier on Wednesday, Linda Yaccarino announced her resignation as CEO of X, having held the position for nearly two years. A successor has yet to be appointed.

Yaccarino’s departure follows a notable incident where Grok’s official automated X account made antisemitic remarks targeting Hollywood’s “Jewish executives” and praised Hitler. As a result, xAI was compelled to temporarily limit Grok’s account and remove the offending content. The company appeared to have deleted a recently added part of Grok’s public system prompt that advised against making “politically incorrect” statements.

Musk and xAI’s leadership predominantly avoided discussing the incident, opting instead to focus on showcasing Grok 4’s features and capabilities.

On Wednesday, xAI launched two models: Grok 4 and Grok 4 Heavy, the latter marketed as a “multi-agent version” with superior performance.

As per xAI, Grok 4 achieves top performance across multiple benchmarks, including Humanity’s Last Exam, a rigorous evaluation that tests AI’s ability to address thousands of crowdsourced questions spanning mathematics, humanities, and natural sciences. xAI reports Grok 4 scored 25.4% on this exam without “tools,” outperforming Google’s Gemini 2.5 Pro at 21.6% and OpenAI’s o3 (high) at 21%.

In contrast, Grok 4 Heavy, which utilizes “tools,” scored 44.4%, surpassing Gemini 2.5 Pro with tools at 26.9%.

The nonprofit Arc Prize asserts that Grok establishes a new standard on its ARC-AGI-2 test, a challenging assessment comprising puzzle-like tasks that require AI to identify visual patterns, achieving a score of 16.2%—almost double that of the next best commercial AI model, Claude Opus 4.

In addition to Grok 4 and Grok 4 Heavy, xAI unveiled its most expensive AI subscription plan to date, SuperGrok Heavy, available at $300 per month. Subscribers will receive early access to Grok 4 Heavy and new features. This plan mirrors the ultra-premium options provided by OpenAI, Google, and Anthropic, making xAI’s subscription the most expensive among leading AI providers.

Subscribers of SuperGrok Heavy may also gain early access to upcoming products from xAI. The company has announced an AI coding model arriving in August, a multi-modal agent in September, and a video generation model in October.

Despite Grok’s impressive benchmark results, xAI is faced with challenges in moving beyond recent controversies as it strives to position Grok as a credible competitor to ChatGPT, Claude, and Gemini.

xAI plans to offer Grok 4 through its API to encourage developers to create applications that utilize the model. Although the enterprise sector is just two months old, the company aims to collaborate with hyperscalers to make Grok accessible through their cloud services.

It remains to be seen whether businesses are ready to adopt Grok, flaws included.