GPT-4.5 six-hour top-ranking battle: Grok-3 hits the top

OpenAI's GPT-4.5 model successfully topped the artificial intelligence arena within six hours after its release and became the first in the entire task category. However, this glory did not last long. Musk's xAI Grok-3 model quickly counterattacked and surpassed it to become the first place in the overall list.

According to voting data, GPT-4.5 and Grok-3 each received more than 3,000 votes, with the final total score of 1,412 vs. 1,411, only one point away. Although GPT-4.5 performs well on most projects, Grok-3 has a slight advantage in specific "style control" and "difficult prompt words" tasks, which makes it a turnaround in total score.

Regarding this "six-hour reversal", many users expressed doubts about whether such a rapid change was reasonable. In response, some industry insiders explained that there is a voting threshold for the competition list, and only a model with 3,000 votes can be on the list at the same time. Therefore, it is a coincidence that these two models, which have just been released, can meet this standard simultaneously.

It is worth mentioning that although GPT-4.5 faced some negative reviews when it was first released, users' recognition of its high emotional intelligence has increased significantly in the future. OpenAI CEO Sam Altman even shared a conversation with GPT-4.5, saying it was the first time he had received a request from users that he promised not to remove the model.

Meanwhile, GPT-4.5 also performed well in an alternative competition, participating in a game similar to "Mobile Werewolf Kill". In this game, major AI models need to be debated, strategy formulation and voting, and the final winner is decided by a jury composed of eliminated members. GPT-4.5 has shown outstanding performance beyond humanity in cooperation, deception and strategy formulation.

All this shows that the competition for artificial intelligence is becoming increasingly fierce, and major models are constantly innovating and improving in their respective fields. In the future, who will eventually win this smart battle is worth our continuous attention.