Grok 3 Unveiled: xAI’s Bold Bid for AI Supremacy Takes Center Stage
Elon Musk Claims New AI Outsmarts Rivals, Signals a New Era for xAI’s Ambitions February 18, 2025 SAN FRANCISCO — In a livestreamed event on X late Monday night, Elon Musk and his team at xAI pulled back the curtain on Grok 3, a new artificial intelligence model they boldly
Elon Musk Claims New AI Outsmarts Rivals, Signals a New Era for xAI’s Ambitions
February 18, 2025
SAN FRANCISCO — In a livestreamed event on X late Monday night, Elon Musk and his team at xAI pulled back the curtain on Grok 3, a new artificial intelligence model they boldly claim is "the smartest AI on Earth." Unveiled on February 17, 2025, Grok 3 arrives with promises of unmatched reasoning, computational power, and a mission to seek truth above all—positioning xAI as a formidable challenger to AI giants like OpenAI, Google, and China’s DeepSeek.
The launch, watched by millions on X, showcased a model that Musk says leapfrogs its predecessor, Grok 2, and competitors alike. Trained on a staggering "Colossus" supercomputer cluster packing 200,000 Nvidia GPUs in Memphis, Grok 3 underwent a grueling two-phase training process—122 days on 100,000 GPUs, then 92 days scaled to the full cluster. The result, per xAI, is a system with over ten times the compute power of Grok 2, capable of tackling complex math, science, and coding challenges with ease.
Grok-3 is the first model *ever* to score over 1400 on Chatbot Arena and outperforms the best publicly available reasoning models from OpenAI and Google.
— Gavin Baker (@GavinSBaker) February 18, 2025
xAI was founded 13 years after Deepmind and 8 years after OpenAI and is now ahead of both. The “SR-71 Blackbird” of AI labs. pic.twitter.com/ExcdENPu4U
"Grok 3 isn’t just an upgrade—it’s a paradigm shift," Musk declared during the event, calling it "scary smart." He pointed to benchmark victories, including a 1,400 Elo score on Chatbot Arena—the first model to hit that mark—along with top-tier results in AIME (math), GPQA (science), and LiveCodeBench (coding). Even a slimmed-down version, Grok 3 Mini, reportedly holds its own against industry leaders.
The AI’s capabilities shone in live demos: plotting a Mars mission, designing a Tetris-Bejeweled hybrid game, and fielding philosophical queries like "What’s the meaning of life?" Two reasoning modes—“Think” for transparent step-by-step logic and "Big Brain" for heavy-duty problem-solving—underscored its versatility. Musk emphasized its "maximally truth-seeking" design, a dig at what he’s called the "woke" tendencies of rival models.
Alongside Grok 3, xAI debuted "Deep Search," a research-driven search engine tapping X posts and web data for detailed insights. A voice mode is also imminent, expected within days, with broader API access and Tesla integration on the horizon. For now, Grok 3 rolls out to U.S.-based X Premium Plus subscribers ($40/month) and a new "SuperGrok" tier ($30/month), with Grok 2 set for open-sourcing once its successor stabilizes.
Grok 3 was just released. You won't believe it, I've already created a game.
— Penny2x (@imPenny2x) February 18, 2025
(I got early access THIS MORNING).
This game was 100% created by GROK, I just told it what I wanted, and put the code in the right place.
I just keep asking for adjustments, and it keeps spitting… pic.twitter.com/BMtIe3U4KF
The unveiling escalates Musk’s rivalry with OpenAI, where he co-founded the group before a bitter split, and follows his rejected $97.4 billion buyout offer last month. xAI’s claims pit Grok 3 against GPT-4o, Claude 3.5 Sonnet, Gemini 2.0 Pro, and DeepSeek V3—though independent tests are still pending to verify its edge.
Skeptics abound. While the demos dazzled, some analysts argue Grok 3 catches up to, rather than redefines, the AI frontier. Yet with xAI’s $50 billion valuation potentially climbing to $75 billion via a rumored $10 billion funding round, the company’s momentum is undeniable.Daily updates are promised as Grok 3 evolves from its beta phase, a sign xAI sees this as just the beginning. For Musk, it’s more than a product—it’s a step toward unraveling the universe’s mysteries, one truth-seeking answer at a time.