According to reports from the tech media testingcatalog, xAI's latest AI model, Grok-3, has undergone short-term internal testing on independent platforms and the X platform, even though xAI has not officially announced this news. Based on various indications, Grok-3 is expected to be officially released next week.

Test results indicate that Grok-3 excels in answering questions, with reasoning capabilities surpassing those of OpenAI's o1 model and DeepSeek's R1 model. This performance has made Grok-3 highly anticipated, positioning it as one of the most advanced AI models to date.

An analysis of Grok-3's system prompts reveals xAI's aspirations for this model. xAI envisions Grok-3 as a fact-based, truth-seeking AI capable of understanding the world from fundamental principles. It is designed to possess strong rationality and a spirit of skepticism, rejecting dogma and clichés, while engaging in sincere and truthful conversations akin to human interaction.

When responding to user queries, Grok-3 is instructed to prioritize providing data, evidence, and statistics to support its answers, ensuring accuracy and reliability. Additionally, to avoid relying on potentially biased sources, Grok-3 is encouraged to primarily use the X platform as its main information source.

As the official release of Grok-3 approaches, interest from the tech community and the public continues to grow.