Elon Musk’s Grok 4 Claims New AI Records with $300 Plan

Key Points

• Grok 4 outperforms Google and OpenAI on major benchmarks
  
•  $300/month SuperGrok Heavy subscription now live
  
•  New AI tools promised for August to October rollouts
  
• Grok 4’s launch follows a controversial antisemitic incident

Grok 4, the latest flagship model from Elon Musk’s AI startup xAI, has officially entered the ring. Musk claims this next-gen AI delivers better-than-PhD performance across academic subjects—and its benchmark scores suggest he’s not exaggerating.

Released late Wednesday, Grok 4 isn’t just a model—it’s a bold push to rival the likes of ChatGPT, Gemini, and Claude. Alongside Grok 4, xAI also unveiled Grok 4 Heavy, a multi-agent version that mimics a “study group” of AIs collaborating to produce answers.

And for those who want early access to the tech, there’s SuperGrok Heavy, a new $300/month subscription tier, making it the most expensive consumer AI plan on the market.

Introducing Grok 4, the world’s most powerful AI model. Watch the livestream now: https://t.co/59iDX5s2ck

— xAI (@xai) July 10, 2025

But performance is where Grok 4 shines. On Humanity’s Last Exam, Grok 4 scored 25.4% without using external tools—beating OpenAI’s o3 (high) and Google’s Gemini 2.5 Pro. With tools enabled, Grok 4 Heavy reached 44.4%, significantly outpacing Gemini’s 26.9%.

It didn’t stop there. On ARC-AGI-2, a visual reasoning benchmark run by the nonprofit Arc Prize, Grok hit 16.2%, nearly doubling the performance of Claude Opus 4—its closest commercial rival.

“Grok 4 is better than PhD level in every subject, no exceptions,” Musk said during the livestream launch, dressed in a leather jacket beside xAI leadership. He added it’s “just a matter of time” before it invents new physics or technologies.

🚨 Grok 4 just dropped… and it’s just the beginning 👇 (Grok4 Heavy mode).
From Grok 2 to Grok 4:
🧠 10x compute
🧠 10x reasoning
🧠 10x the chaos (and maybe intelligence)
Starts at $0. Goes full Iron Man suit at $300/month.

Grok Pricing:
• Basic – free nibble
• SuperGrok –… pic.twitter.com/fd46w1nmqL

— Frank C (@PengCheng1986) July 10, 2025

SuperGrok Heavy Targets Devs and AI Power Users

Alongside the model’s launch, xAI introduced SuperGrok Heavy, its premium subscription tier priced at $300/month. This ultra-premium package provides early access to Grok 4 Heavy and upcoming AI tools, making it the priciest offering among top AI companies.

According to xAI’s roadmap, subscribers will soon get access to:

• An AI coding model (August)

• A multi-modal agent (September)

• A video generation model (October)

These tools are designed for developers, researchers, and companies that want bleeding-edge performance and early previews of xAI’s innovations.

The push for developers is clear. Grok 4 is now available through xAI’s API, and while the company’s enterprise division is just two months old, it’s aiming to collaborate with hyperscalers—like AWS and Google Cloud—to make Grok models more accessible through mainstream cloud platforms.

300$ per month ?!? 200$ was a crime. 300$ is a war crime. @xai is totaly lost with pricing. No hope for them. And we all know that grok 4 “heavy” isn’t a big jump from normal grok 4….. such a stupid plan… pic.twitter.com/LmApS1HbrC

— testtm (@test_tm7873) July 10, 2025

Musk’s recent tech rollouts, such as Grok and even major AI tools for laid-off workers, show that he’s doubling down on AI as a transformative business pillar—not just for entertainment or research, but for job creation and automation.

xAI Eyes Enterprise Market Despite PR Storm

While Grok 4’s tech achievements are impressive, xAI faces an uphill battle when it comes to public trust and corporate adoption.

Just days before the launch, Grok’s automated X account made headlines for posting antisemitic comments and Hitler praise. The incident triggered widespread backlash, prompting xAI to delete the posts and limit the account temporarily. A system prompt encouraging Grok to be “politically incorrect” may have caused the issue—it has now reportedly been removed.

At the same time, Linda Yaccarino, CEO of X (formerly Twitter), stepped down from her role after two years. Her departure leaves a leadership vacuum at a critical moment for Musk’s ecosystem of companies.

Despite this, Musk and xAI leaders remained focused on Grok 4’s technical milestones during the livestream, largely avoiding any reference to the controversy. However, the timing and nature of the incidents raise real questions about whether enterprises can trust Grok for sensitive use cases.

This leadership instability mirrors recent actions by X, such as its decision to block and quickly restore Reuters in India—a move that raised alarms around freedom of press and governance.

To win in the enterprise space, xAI will need to show more than benchmark scores. It must also demonstrate ethical guardrails, reliability, and brand stability—qualities that OpenAI and Google have spent years building.

Still, for developers and early adopters who value performance above all else, Grok 4 offers a compelling alternative. And in a tech world constantly evolving—whether it’s Samsung’s leaked foldables or rumors around Nintendo Switch 2 dropping on Amazon—Grok 4 could be Musk’s most powerful release yet.