OpenAI officially launches o3-mini

The o3-mini is OpenAI's first free-to-use reasoning model. It is designed to provide higher accuracy than conventional large language models. It uses a system of "thinking sequences" before coming up with an answer and correcting itself. This makes it ideal for the most difficult questions, especially in the field of mathematics.

o3-mini can compete with leading models like o1 and o1-mini at just $1.10 per million input tokens - half that of o1-mini.

However, compared to the DeepSeek reasoning model, the cost of o3-mini is still double - reports reveal that DeepSeek's R1 model costs only $0.60 per million input tokens.

The o3-mini's launch was unexpected by the tech community, possibly due to OpenAI's decision to accelerate the release to accommodate DeepSeek's growth.

Unlike DeepSeek, o3-mini is not open source and free users do not have unlimited access. To use o3-mini, users need to press the "reason" button before sending a question to ChatGPT. This option is now available on the ChatGPT website.

Source: https://kinhtedothi.vn/openai-chinh-thuc-ra-mat-o3-mini.html