OpenAI announced the full version of its reasoning model and a new ChatGPT subscription tier with access to “the smartest model in the world.”
The artificial intelligence startup said its o1 model is now out of preview and available through ChatGPT Plus, during the first day of its “12 Days of OpenAI” event on Thursday. ChatGPT Team users also have access to o1 out of preview, while Enterprise and Edu users will get access to the full model next week.
OpenAI chief executive Sam Altman was joined by some OpenAI team members at the event where they said one of the main complaints with o1-preview was that it is “way too slow.”
o1, which has been in preview since September, now has a faster response time and “more powerful reasoning” capabilities that make it better for coding, math, and writing tasks, the startup said. According to testing published by OpenAI, o1 made 34% fewer mistakes than the preview.
The full version can also support image inputs, “allowing it to apply reasoning to visuals for more detailed & useful responses.” OpenAI said it would add o1 to its API for developers soon, and that it is planning to announce more multimodal features in the future.
In addition to o1's full release, OpenAI announced a new, $200-a-month subscription tier for ChatGPT called ChatGPT Pro. Through ChatGPT Pro, users can have unlimited access to the startup’s AI models and tools, including a more advanced version of o1 called o1 pro mode and Advanced Voice.
The pro version of o1 uses more compute power to “think harder and provide even better answers,” the startup said. ChatGPT Pro will eventually offer more compute-intensive productivity features.
In September, OpenAI announced the OpenAI 01 model series, which it said is “designed to spend more time thinking before they respond.” The company also announced OpenAI o1-mini, a smaller and cheaper version of the model, which can help developers with coding tasks.