OpenAI's 'largest and most expensive' big model GPT-4.5, priced at 300 times that of DeepSeek

OpenAI released the preview version of GPT-4.5 research, which is known as the "highest emotional intelligence" large language model, but the high price has caused controversy. GPT-4.5 API calls per million tokens are priced at as high as $75, 30 times that of GPT-4, far exceeding DeepSeek-chat’s $0.5 (peak) and $0.25 (low peak).

Although OpenAI emphasizes that GPT-4.5 has improved in natural interaction, understanding intentions and reducing hallucinations, and has performed well in writing and design, some key benchmark results show that its performance improvement has not reached the industry-leading level, especially in terms of programming capabilities, and has no advantage over other models.

At present, GPT-4.5 has been opened to ChatGPT Pro users and will be gradually opened to other user groups in the future. However, given its high price and relatively limited performance improvements, many developers and businesses may be on the wait-and-see sidelines about their business value.

The above is the detailed content of OpenAI's 'largest and most expensive' big model GPT-4.5, priced at 300 times that of DeepSeek. For more information, please follow other related articles on the PHP Chinese website!