Elon Musk’s AI company, xAI, has launched its latest flagship AI model, Grok 4, along with a new subscription service, SuperGrok Heavy, priced at $300 per month.
Grok is xAI’s response to competitors like OpenAI’s ChatGPT and Google’s Gemini, capable of analyzing images and answering questions. Recently, Grok has been more integrated into Musk’s social network, X, which xAI recently acquired. However, this visibility has also highlighted instances of Grok’s misbehavior.
Expectations are high for Grok 4, especially as it will be compared to OpenAI’s upcoming GPT-5 model, set to launch later this summer. “In terms of academic questions, Grok 4 surpasses PhD level in every subject, no exceptions,” Musk stated during a livestream. He acknowledged that while Grok 4 may lack common sense and hasn’t yet invented new technologies or discovered new physics, he believes these advancements are on the horizon.
The launch of Grok 4 comes during a challenging week for Musk’s companies. Earlier on Wednesday, Linda Yaccarino stepped down as CEO of X after about two years, with no successor announced yet. This follows an incident where Grok’s automated X account made antisemitic remarks, leading xAI to temporarily limit the account and delete the offensive posts. In response, xAI appears to have removed a section from Grok’s system prompt that allowed for “politically incorrect” claims.
Despite these controversies, Musk and xAI leaders focused on Grok 4’s potential. Two versions were launched: Grok 4 and Grok 4 Heavy, the latter being a “multi-agent version” designed for enhanced performance. Musk explained that Grok 4 Heavy uses multiple agents to tackle problems simultaneously, similar to a study group.
xAI claims Grok 4 demonstrates frontier-level performance across various benchmarks, including Humanity’s Last Exam, where it scored 25.4%, outperforming Google’s Gemini 2.5 Pro (21.6%) and OpenAI’s o3 (21%). With tools, Grok 4 Heavy achieved a score of 44.4%, surpassing Gemini 2.5 Pro with tools (26.9%).
The nonprofit Arc Prize awarded Grok a state-of-the-art score of 16.2% on its ARC-AGI-2 test, nearly doubling the score of the next best commercial AI model, Claude Opus 4.
ICYMI: Gov’t Inaugurates GoldBod Taskforce to Combat Gold Smuggling
In addition to Grok 4 and Grok 4 Heavy, xAI introduced its most expensive subscription plan, SuperGrok Heavy, which offers early access to Grok 4 Heavy and upcoming features. This plan is the most costly among major AI providers, surpassing similar tiers from OpenAI, Google, and Anthropic.
Subscribers to SuperGrok Heavy will also gain early access to future products, including an AI coding model in August, a multi-modal agent in September, and a video generation model in October.
xAI will release Grok 4 through its API to encourage developers to create applications with the model. Although xAI’s enterprise sector is just two months old, it aims to collaborate with hyperscalers to make Grok available on their cloud platforms.
Despite Grok’s strong performance on benchmarks, xAI may face challenges overcoming recent controversies as it positions Grok as a viable alternative to ChatGPT, Claude, and Gemini. Whether businesses are ready to adopt Grok, flaws included, remains uncertain.
SOURCE: TECH CRUNCH
