According to Reuters on February 28, artificial intelligence company OpenAI announced the launch of its latest large-scale language model, GPT-4.5 research preview. The model is called by OpenAI its "largest and most knowledge-rich" chat model, with stronger pattern recognition capabilities, higher emotional intelligence and lower hallucination rate.

However, despite the excellent performance of GPT-4.5, its huge model size and high computing cost also bring challenges. OpenAI CEO Sam Altman said that due to the shortage of GPU resources, the promotion rhythm has been affected to a certain extent. He revealed that OpenAI will add tens of thousands of GPUs next week to support wider user access.

In addition, the API pricing of GPT-4.5 is relatively high, with the price of input tokens being 30 times that of GPT-4o and the price of output tokens being 15 times. OpenAI is evaluating whether to provide this service in the API for a long time.