TECH NEWS – xAI, Elon Musk’s AI-focused company, has unveiled its latest large language model (LLM).

In a live broadcast (which you can watch below), they demonstrated Grok 3, which is available to Twitter users who subscribe at the most expensive level, as it is available in the Premium+ category. While the artificial intelligence company continues to tout the new LLM’s capabilities as best-in-class, some experts point to critical shortcomings in the benchmarks released. Musk announced that the older Grok 2 LLM will be open-sourced in a few months.

The xAI was keen to note that Grok 3 LLM beat all other publicly released versions of the base model, including DeepSeek-V3 and GPT-4o, in math, science and coding benchmarks. The LLM achieved an unprecedented score of 1402 on the Arena benchmark. Meanwhile, Manifold Markets’ bet that Grok 3 is the world’s most powerful artificial intelligence is now expected to receive an overwhelming majority of yes votes. It should be added, however, that the probability of a yes win has dropped from 91 percent late Monday night to just 78 percent. Critical comments around xAI Grok 3 may have played a role in this.

Zihan Wang (who coincidentally also used to work at DeepSeek) showed Grok 3 a picture of two different sized iron balls hanging from the Leaning Tower of Pisa at different heights, and then asked which ball would land first. The logical answer could only be the one that was heavier and closer to the ground, but LLM replied that both balls would land at the same time. Others asked why xAI did not publish the Grok 3 score on the FrontierMath, Arc-AGI, or HLE benchmarks.

This may raise the question of whether it is really the best LLM in its category. Meanwhile, Bloomberg recently reported that xAI is in talks with existing investors to raise up to $10 billion in a new funding round that would value the startup at $75 billion. In its last funding round, xAI raised $6 billion at a valuation of $40 billion. Guodang Zhang of xAI confirmed that Grok 3 has been trained on 100,000 GPUs.

You’d have to back up that claim, wouldn’t you, Elon?

Source: WCCFTech

GROK 3: SOLVING PHYSICS, GAMES, AND THE UNIVERSE

Full presentation and demo of xAI’s latest model

0:00 xAI’s mission: Understand the universe
1:20 Team presentation
2:01 Grok means to profoundly understand
2:29 From Grok 2 to Grok 3
6:30 Grok 3 benchmarks
9:07 Grok 3 improves… https://t.co/7qbB6O16Yb pic.twitter.com/BomGwAOa1I

— Mario Nawfal (@MarioNawfal) February 18, 2025