technology

OpenAI Introduces More Advanced Reasoning AI Models

OpenAI has launched a more advanced version of its o1 reasoning AI model in its developer API. 

According to a TechCrunch report on Wednesday (3/19), the reasoning AI model, called o1-pro, is claimed to use more computing power than the previous version to produce consistently better responses. 

Currently, the o1-pro model is only available to certain developers who have spent at least $5 (around Rp. 82.6 thousand) in the OpenAI API service and the price is considered very expensive. 

OpenAI charges $150 (around Rp. 2.4 million) per million tokens (around 750,000 words) of input and $600 per million tokens of output generated by the model. 

The cost of using the o1-pro model is twice as large as GPT-4.5 for input and ten times more expensive than the regular o1 model. 

OpenAI is betting that the increased performance of o1-pro will convince developers to pay a premium for the service. 

An OpenAI spokesperson said the o1-pro model was designed to “think harder” and provide better answers to the toughest problems. 

“After receiving a lot of requests from the developer community, we are excited to bring this model into the API to offer more reliable responses,” he said. 

However, the initial response to the o1-pro model has not been entirely positive. 

The model, which has been available on OpenAI’s ChatGPT platform for ChatGPT Pro customers since December 2024, has reportedly had trouble solving Sudoku puzzles and been fooled by simple optical illusion-based jokes. 

Additionally, OpenAI’s internal benchmarks from late last year showed that o1-pro performed only slightly better than the standard o1 on programming and math problems. However, it did provide more reliable answers.?????? 

Tag: ChatGPT OpenAI