Below is a summary of announcements made by OpenAI on the 6th of November 2023
- GPT-4 Turbo: A new and improved version of the GPT-4 model that is more capable, cheaper, and supports a larger context window.
- Assistants API: A new API that makes it easier for developers to build their own assistive AI apps that have goals and can call models and tools.
- Multimodal capabilities: New features that enable the platform to handle vision, image creation, and text-to-speech tasks.
- Price reductions: A significant decrease in the operational costs of GPT-4 Turbo and other models and tools.
GPT-4 Turbo
GPT-4 Turbo is the next generation of the GPT-4 model that was released in March 2023 and made generally available in July 2023. GPT-4 Turbo is more capable and has knowledge of world events up to April 2023. It has a 128k context window so it can fit the equivalent of more than 300 pages of text in a single prompt. It also has improved performance and accuracy, especially for tasks that require following instructions or generating specific formats. GPT-4 Turbo is available for all paying developers to try by passing gpt-4-1106-preview in the API and the stable production-ready model will be released in the coming weeks. GPT-4 Turbo is also 3x cheaper for input tokens and 2x cheaper for output tokens compared to GPT-41.
Assistants API
Assistants API is a new API that makes it easier for developers to build their own assistive AI apps that have goals and can call models and tools. Assistants API allows developers to define functions of their app or external APIs and have the model intelligently choose to output a JSON object containing arguments to call those functions. Assistants API also supports calling multiple functions in a single message, such as “open the car window and turn off the A/C”. Assistants API includes several built-in assistants, such as Code-Interpreter, which can execute code snippets in various languages, and Retrieval, which can fetch information from the web or other sources.
Multimodal capabilities
Multimodal capabilities are new features that enable the platform to handle vision, image creation, and text-to-speech tasks. Vision is a new model that can perform image classification, object detection, face recognition, and other vision tasks. Image creation is a new feature that allows developers to use DALL·E 3, the latest version of the image synthesis model that can create images from text or sketches. Text-to-speech is a new feature that allows developers to use TTS, a model that can generate realistic human speech from text.
Price reductions
Price reductions are a significant decrease in the operational costs of GPT-4 Turbo and other models and tools. Input tokens are now 3x cheaper and output tokens are 2x cheaper for GPT-4 Turbo compared to GPT-4. Function calling is now free for all developers. Vision, image creation, and text-to-speech are also offered at competitive prices. These price reductions are part of OpenAI’s mission to democratize AI technology and make it more accessible to a broader range of users and applications.
You can read their full release here – https://openai.com/blog/new-models-and-developer-products-announced-at-devday
