Back in September, OpenAI announced the new o1 series of AI models designed to spend more time thinking before they respond. These models can reason through complex tasks and solve harder problems than previous OpenAI models in science, coding, and math. ChatGPT paid plan users already have access to the o1 preview models.
Today, OpenAI announced that the o1 model is now out of preview. Compared to the o1 preview, the new o1 model will be faster and it will perform better at coding, math & writing. Also, the o1 model now supports image uploads. This will allow users to upload images and get detailed responses back from the model.
OpenAI claims that the new o1 is more concise in its thinking, resulting in faster response times than o1-preview. According to their own internal benchmarks, o1 outperforms o1-preview, reducing major errors on difficult real-world questions by 34%. OpenAI also released the updated o1 system card. As expected, the o1 model still has a "medium" risk rating based on testing with an expanded suite of evaluations.
The full o1 model will be available for ChatGPT Plus and Team users today, and it will replace the existing o1-preview model. ChatGPT Enterprise and Edu users will have access to the o1 model in one week.
During the launch live stream, the o1 team mentioned that they are still working on adding support for tools like web browsing and file upload in ChatGPT. Also, developers will have to wait a little longer to access o1 via APIs. The OpenAI o1 team is still working to bring the o1 API with support for function calling, developer messages, structured outputs, and vision.
Today, OpenAI also announced the new ChatGPT Pro plan that will have access to o1 pro mode, which uses more compute for the best answers to the hardest questions. Compared to both o1 and o1-preview, o1 pro mode will perform better on challenging ML benchmarks across math, science, and coding. The ChatGPT Pro model will cost $200 per month and you can learn more about it here.