xAI Grok 3 overtakes DeepSeek on App Store: Becomes second most popular chatbot within 12 hours of launch

Elon Musk’s AI chatbot, Grok 3, has surged past DeepSeek to claim the number two spot among free apps on Apple’s U.S. App Store. This leap happened in just 12 hours after its successful launch, thanks in part to Musk’s much-hyped description of Grok 3 as “the smartest AI on Earth.”

App Store data shows that Grok 3 has outpaced China’s DeepSeek but still sits behind OpenAI’s ChatGPT, which continues to dominate as the world’s top chatbot.

Grok 3 Fails the Hexagon Ball Bouncing Test

Not everything is smooth sailing for Grok 3. Despite its rapid climb in downloads, the chatbot struggled with coding tasks, trailing OpenAI’s 03-mini and DeepSeek R1 models. One notable failure: the ‘hexagon ball bouncing’ test, where it was asked to write a Python program for a ball bouncing inside a spinning hexagon.

“Grok 3 is here and it fails the ‘hexagon ball bouncing’ test spectacularly,” a user named Theo shared on X.

Grok 3 is here and it fails the “hexagon ball bouncing” test spectacularly pic.twitter.com/548qXruFMv

— Theo – t3.gg (@theo) February 18, 2025

Musk and the xAI team introduced Grok 3 yesterday at 8 p.m. Pacific Time (11 p.m. Eastern), giving the public its first look at what Musk described as the ‘smartest AI on Earth.’ In a demo posted on X, Grok 3 showed impressive results in math, science, and coding benchmarks, outperforming competitors like Gemini-2 Pro, Claude 3.5 Sonnet, and GPT-4o in key performance tests.

The primary Grok 3 model is being rolled out gradually through the Grok app, while a smaller ‘Grok 3 mini’ version offers faster response times.

Both models have secured top spots on the AIME‘24, GPQA, and LiveCodeBench leaderboards, with Grok 3 landing the number one position on Chatbot Arena. They also come with ‘reasoner’ variations designed to solve problems through more analytical thinking, similar to OpenAI’s o3-mini and DeepSeek R1.

Training Grok 3 took ten times more computing power than its predecessor, Grok 2, with xAI relying on its Colossus supercomputer—a system equipped with 200,000 H100 GPUs.

Why It Matters

The success of Grok 3 positions xAI, just two years old, as a key contender in the AI space. Still, the competition isn’t standing still. OpenAI is preparing to launch GPT-4.5 and, eventually, GPT-5. Meanwhile, Anthropic, DeepMind, and Chinese tech giants like Alibaba and DeepSeek are all working to secure their share of the AI market. The race is heating up, and Grok 3’s rapid ascent is just the beginning.

According to xAI, Grok 3 was trained with synthetic data, which Musk says helps it go back over its own work to spot and fix mistakes. This self-correcting behavior might give it an edge over other models that sometimes double down on errors.

You’ll be able to access Grok 3 through Musk’s social platform X, formerly known as Twitter, and possibly through its own app. xAI is pushing to establish itself in the crowded AI market, and Grok 3 is a big part of that strategy.

Behind the scenes, xAI is reportedly trying to raise $10 billion, which would give the company a valuation of about $75 billion. The plan is to pour that money into expanding data center capacity and refining their AI models. Musk has made it clear that xAI isn’t just here to play around. His goal is to build AI tools that focus on truth and transparency, standing apart from competitors like OpenAI and Google.

Late last year, xAI has closed a massive $6 billion Series C funding round. This latest investment includes high-profile names like A16Z, Blackrock, and Fidelity Management & Research Company.

With the demo happening tonight, expectations are sky-high. Will Grok 3 live up to the hype and prove it’s more than just big talk? We’ll find out soon enough.

Musk founded xAI in 2023 and later introduced Grok as a direct challenger to OpenAI’s ChatGPT.