Tech mogul Elon Musk’s AI startup, xAI, has introduced Grok, a new AI chatbot that’s giving ChatGPT a run for its money. What sets Grok apart? Well, it seems to have a bit of a “rebellious streak.” According to xAI’s blog post, Grok takes inspiration from Douglas Adams’ famous science fiction comedy, “The Hitchhiker’s Guide to the Galaxy.” Grok is currently in beta testing and has big plans for growth in the coming months. While Grok and ChatGPT both engage in conversations and answer questions, let’s dive into the five key distinctions that set them apart.
Grok is powered by Grok-1, an advanced LLM (Large Language Model) developed by xAI with a whopping 33 billion parameters. ChatGPT, on the other hand, is built upon OpenAI’s GPT model. Grok’s primary goal, according to xAI, is to help people understand and learn. It’s worth noting that Grok-1 LLM, created over four months, serves as the engine behind Grok.
Grok-1 boasts benchmark performance, with a score of 73% on MMLU (Middle Math-Logic-Understanding) and 63.2% on the HumanEval coding challenge. While Grok-1’s performance has rapidly improved compared to Grok-0, it still falls short of GPT-4. In specific math tests, Grok-1 outperforms GPT-3.5 but lags behind GPT-4, Palm 2, Claude 2, and GPT-3.5. It’s important to note that Grok-1 achieves these results even with a smaller dataset and less processing power, as stated by xAI.
3. Hand-Graded Grok
xAI conducted a hand-grading evaluation, where Grok-1 secured a C grade (59%) in the 2023 Hungarian national high school finals in mathematics, surpassing Claude 2, which scored 55%, and coming close to GPT-4’s B grade (68%). This clear comparison demonstrates that Grok-1 is already more powerful than OpenAI’s GPT-3.5 but not as advanced as the latest GPT-4 model.
ChatGPT offers two versions: basic (free) and premium ($20 per month). The premium version provides real-time access. In contrast, Grok comes at a monthly cost of $16 for X’s premium+ subscribers.
A significant distinction is that all premium users of X, formerly known as Twitter, will gain access to this AI chatbot once it’s out of testing. Grok will provide real-time access to the X platform. However, it’s essential to note that the AI chatbot’s early access will be limited to verified users on X, as indicated on the official Grok webpage.
In conclusion, Grok and ChatGPT may share some common functions, but these key differences set them apart in terms of purpose, capabilities, grading, pricing, and accessibility.