Grok 3 is here to take on OpenAI and DeepSeek
If Grok 2 was still learning, Grok 3 is now “college-ready,” as Musk put it.
Elon Musk has been talking a big game about AI for years, but let’s be real—Grok models haven’t exactly shaken up the industry. That might be changing, though, with Grok 3, which its parent company xAI says is 10 times more powerful than its predecessor, Grok 2.
Looking at the benchmark results, Grok 3 also seems to be throwing some serious punches. xAI claims it outperforms GPT-4o, Claude 3.5, and Gemini 2.0 Pro on AI evaluation tests like AIME 2024 and GPQA Science.
Over at LMSYS Chatbot Arena, where AI models go head-to-head in user-voted matchups, Grok 3 (under the alias “chocolate”) became the first model to cross the 1,400 Elo rating mark. If accurate, this could be the moment xAI shifts from Musk’s experimental side project to an actual contender in the AI arms race.
However, xAI isn’t just flexing on benchmarks—it’s also doubling down on reasoning models. The Grok 3 Reasoning model is designed for deeper logical processing, with xAI claiming it surpasses OpenAI's o3-mini and DeepSeek R1. There’s also DeepSearch, an AI agent that scours the web for real-time information, making it xAI’s answer to OpenAI and Google's research-powered AI. And for those looking for even more brainpower, Grok 3's Big Brain mode is expected to tackle even more complex queries.
But of course, there’s always a catch.
Grok 3’s best features aren’t exactly free. The base model is available to X’s Premium+ subscribers, but if you want DeepSearch, Big Brain mode, and unlimited image generation, you’ll need to shell out $30/month for SuperGrok—a new subscription plan. The model is available on iOS and web apps.
And then there’s the timing. Grok 3’s launch comes just days after OpenAI rejected Musk’s $97.4 billion bid to buy the company. Coincidence? Maybe. Personally, though, I think Grok 3 may easily be xAI's most serious attempt yet. Whether it'll finally make xAI a real challenger remains to be seen. Until then, the AI showdown continues.
Emmanuel Oyedeji
