OpenAI's GPT-4.5 model quickly topped the artificial intelligence arena within just six hours after its release and became the champion of the full task classification. This achievement not only demonstrates its strong technical strength, but also attracts widespread attention from the industry. However, this glory did not last long. Musk's xAI Grok-3 model achieved a counterattack in a short period of time and successfully overtook it and became the first in the overall list.
According to the voting data, GPT-4.5 and Grok-3 each received more than 3,000 votes in support, with the final total score of 1412 vs. 1411, only one point apart. Although GPT-4.5 performed well in most tasks, the Grok-3 had a slight advantage in specific tasks such as “with style control” and “difficult prompt words”, which led to a reverse over the overall score. This result not only reflects the expertise of the two models in different fields, but also reflects the diversity and competitiveness of artificial intelligence technologies.
Regarding this "six-hour reversal", many users questioned whether such a rapid change was reasonable. In response, industry insiders explained that the competition list has a strict voting threshold, and only a model with 3,000 votes can be on the list at the same time. Therefore, it is actually a coincidence that these two models can quickly meet this standard after their release. This explanation not only responds to user questions, but also reveals the operating mechanism behind the list.
It is worth mentioning that although GPT-4.5 faced some negative reviews in the early stages of its release, 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. This feedback not only reflects users' love for GPT-4.5, but also demonstrates its outstanding performance in emotional interactions.
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 development and voting, and the final winner is decided by a jury composed of eliminated members. GPT-4.5 has shown excellent performance beyond humans in cooperation, deception and strategy formulation, which not only demonstrates its multifaceted capabilities, but also provides new ideas for the application of artificial intelligence in complex tasks.
All of this shows that competition in the field of 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. With the continuous advancement of technology, the application scenarios of artificial intelligence will become more extensive and its impact on society will become more far-reaching.