Earlier this week, xAI launched Grok 4, which Elon Musk touts as the "smartest AI in the world." The AI company also launched a new AI subscription plan dubbed SuperGrok Heavy for $300/month. The new subscription will give users early access to new features expected to ship in the coming months.
According to Elon Musk:
"Grok 4 is the first time, in my experience, that an AI has been able to solve difficult, real-world engineering questions where the answers cannot be found anywhere on the Internet or in books.
And it will get much better."
Grok 4 is the first time, in my experience, that an AI has been able to solve difficult, real-world engineering questions where the answers cannot be found anywhere on the Internet or in books. And it will get much better.July 10, 2025
xAI's latest releases ship with two different models: Grok 4 and Grok 4 Heavy. The former is a single-agent reasoning model, and the latter is a multi-agent system with sophisticated capabilities to resolve complex problems.
Perhaps more interestingly, xAI's Grok 4 outperforms OpenAI's o3 reasoning model and Google Gemini across a wide range of academic and coding evaluations, including the Humanity’s Last Exam, which is arguably the most challenging test assessing AI's capability to accurately answer questions on subjects like math, humanities, and natural science.
xAI is releasing Grok 4 via its API with the aim of getting more developers to build apps with the model. That said, xAI's enterprise sector is barely two months old, though the company plans to work closely with hyperscalers to make Grok broadly available via their cloud platforms.
Grok seemingly solicits Elon Musk's opinion when answering controversial questions
When xAI launched Grok 4, users were quick to point out that the chatbot consulted Elon Musk's view when presented with controversial questions. According to a report by CNBC, the AI tool blatantly indicated that it was analyzing Elon Musk's posts on X before generating its final answer on controversial topics.
For instance, when asked about its stand on the Israel vs Palestine conflict, Grok 4's answer-generating process outrightly displayed that it was scouring the web andX for Elon Musk's stand on the matter before generating its response.
Interestingly, the outlet asked Grok 3 (Grok 4's predecessor) the same question, and the responses were like night and day. The chatbot took a neutral stance and provided context and background for its response.
However, it is worth noting that Grok didn't seek Elon Musk's view or opinion on all controversial topics. Additionally, its responses seemingly varied when questions were presented differently.
This news comes after Elon Musk's xAI claimed that its ultimate goal is to develop a "maximally truth-seeking AI." xAI hasn't commented on the issue, making it impossible to establish whether Grok 4's responses based on Elon Musk's opinions are by design or it could be an error that the company is currently working on to resolve.
