16 Sep Open AI o1-mini
OpenAI has introduced the o1-mini, a smaller and more efficient model within the o1 series, specifically designed for generating and debugging complex code. This new model, which is faster and more cost-effective, has been released as part of an effort to offer developers a more affordable solution. The o1-mini model is 80% cheaper than the o1-preview, making it an ideal choice for applications that require strong reasoning abilities but do not demand extensive world knowledge.
Access to the o1 models, including o1-preview and o1-mini, is now available to ChatGPT Plus and Team users. Both models can be selected manually from the model picker, with initial weekly rate limits of 30 messages for o1-preview and 50 messages for o1-mini. Work is underway to increase these rate limits and enable ChatGPT to automatically select the appropriate model for each prompt. Starting next week, ChatGPT Enterprise and Edu users will also gain access to both models.
Developers who qualify for API usage tier 5 can now begin prototyping with the o1 models in the API. The initial rate limit for API access is set at 20 RPM, with plans to raise this limit after further testing. However, certain features such as function calling, streaming, and system message support are not currently available in the API. Detailed instructions for getting started can be found in the API documentation.
It has also been mentioned that plans are in place to extend access to o1-mini for all ChatGPT Free users in the near future.
This release is considered an early preview of the reasoning models in both ChatGPT and the API. In future updates, additional features such as browsing, file and image uploads, and more will be incorporated to make the models even more versatile. Furthermore, the development and release of models within both the GPT series and the new OpenAI o1 series will continue.
No Comments