Elon Musk's xAI Unveils Grok-3: A Powerful New AI Model

Elon Musk's xAI has unveiled Grok-3, a new AI model that promises more power and capabilities. The company showcased a massive computing infrastructure, including a cluster of 200,000 GPUs called "Colossus," which powers Grok-3's training. The development of this infrastructure was more challenging than creating the AI model itself, according to xAI developers. Musk has plans to expand the cluster's capacity to five times its current size, making it the most powerful GPU cluster on earth.

Grok-3 demonstrates impressive performance across standard AI benchmarks. The base model consistently tops the charts in math, science, and coding tests. In blind tests, an early version of Grok-3, codenamed "Chocolate," achieved the best ELO among all LLMs, indicating user preference for its answers. A specialized "Reasoning Beta" variant of Grok-3 pushes math scores even higher, reaching 93% on the AIME 2025 benchmark.

However, when xAI demonstrated Grok-3's capabilities live, the presentation felt more like a game of catch-up than innovation. The team showcased the model solving physics problems and writing game code from scratch, feats already mastered by ChatGPT, Claude, and Google's Gemini. They also introduced DeepSearch, a research agent that scours the web and generates extensive reports on given topics, similar to tools from OpenAI and Google.

Grok-3 is being rolled out slowly, with some enthusiasts already trying it and expressing satisfaction with the results. Computer scientist Lex Friedman praised Grok-3's capabilities, while others compared it to leading market rivals. xAI plans to open-source Grok-2 once Grok-3 is fully mature, continuing its trend of releasing older versions to spur innovation. The real test for Grok-3 will come when xAI rolls out its promised voice features, gaming tools, and API access in the weeks ahead.