Elon Musk succeeded in making ChatGPT lose badly, here is the proof
- A chatbot developed by a company owned by Elon Musk has successfully defeated ChatGPT. In several tests, Grok managed to have a higher score than the chatbot developed by OpenAI.
Grok, which was released some time ago, announced that it had undergone an upgrade with version 1.5. xAI, the chatbot's developer, explains its latest model can understand longer documents, more complex commands and more advanced reasoning.
Entrepreneur noted that Grok 1.5 scored 74.1% on the HumanEval benchmark. It consists of 164 programming problems and is not included in AI model training.
While GPT-4 gets 67%. Chatbot from Google, Gemini Pro has a score of 71.9%.
Grok scored close to Gemini (83.7%) on the MMLU test, for knowledge of 57 elementary to advanced subjects. The latest model received a score of 81.3%, quoted Monday (1/4/2024).
The Mathematics test, which includes competition questions for elementary to high school, recorded Grok getting 50.6%. For this category, the value is still below GPT-4, namely 52.9%.
Musk is known to have big ambitions for his chatbot. In X's account, he said the new Grok will outperform all existing AI in all metrics.
He also said Grok updates will also continue. "Grok 2 is training now," he said.
Announced late last year, Grok is preparing to be available to X users. However, the chatbot can only be used on paid services within the application.
Grok is said to be priced at US$16 per month which includes X premium features.