Elon Musk’s xAI is on the verge of launching Grok 2, an advanced generative AI chatbot that promises to elevate the capabilities of artificial intelligence in real-time interactions.
Set for release soon, Grok 2 builds on the foundation laid by its predecessor, Grok 1.5, and aims to deliver superior performance in various benchmarks, including MATH and HumanEval, which evaluate the AI’s coding and reasoning abilities.
Specifications and capabilities
Grok 2 is designed to be a multimodal AI, meaning it can process both text and visual information. This capability allows Grok 2 to generate images and provide responses based on web search results, significantly enhancing its interaction quality.
The model is inspired by fictional AI characters like JARVIS from “Iron Man,” aiming to provide users with a conversational experience that feels both intuitive and engaging.
One of the standout features of Grok 2 is its ability to access real-time data from X (formerly Twitter), enabling it to deliver up-to-date responses that reflect current events. This feature sets Grok apart from many existing chatbots, which often rely on static datasets.
Additionally, Grok 2 is expected to excel in tasks requiring complex reasoning and understanding of context, making it a valuable tool for both casual users and professionals seeking assistance in various domains.
Training and computational power
The development of Grok 2 required significant computational resources, utilizing approximately 20,000 Nvidia H100 GPUs for its training. This powerful hardware enables the model to process vast amounts of data, enhancing its learning capabilities.
Looking ahead, Grok 3 is anticipated to ramp up this computational demand, with plans to leverage up to 100,000 Nvidia H100 GPUs, marking it as one of the most ambitious AI projects to date.
The infrastructure supporting Grok’s development is being constructed in collaboration with Dell Technologies and Oracle, which are providing the necessary computing power and support.
This partnership is crucial for addressing the high energy demands associated with training large-scale AI models, as the supercomputer powering Grok will consume around 70 megawatts of power—equivalent to the energy needs of a small city.
Availability
Currently, Grok 2 will be accessible primarily to Premium and Premium+ users on X. This tiered access reflects Musk’s strategy to create a more exclusive environment for early adopters while ensuring that the model is refined through user feedback.
Following its initial release, Grok 2 is expected to undergo continuous improvements, leveraging user interactions to enhance its performance and capabilities.
As Grok 2 transitions from beta testing to a full release, it will likely expand its availability to a broader audience. The initial launch will focus on regions including the United States, Canada, Australia, New Zealand, the United Kingdom, India, and several other countries, with plans for wider distribution as the technology matures.
Read About: How to enable/disable Direct Messaging (DMs) on X (Twitter)