OpenAI Developer Conference wows audience with announcement of GPT-4 Turbo and GPT Store.

In the keynote speech at the first OpenAI Developer Conference held in San Francisco, OpenAI announced a new and more powerful version of GPT-4, as well as some specialized generative AI tools called GPTs and an Assistants API to help developers build AI applications. Additionally, OpenAI announced that ChatGPT will use GPT-4 Turbo, which can leverage the latest current affairs information up until April 2023. GPT-4 Turbo offers more capabilities compared to GPT-4, and at a lower price. The new features added to GPT-4 Turbo include: - An increased context length of 128,000 tokens, equivalent to 300 pages of a standard book. - More control, including a JSON schema, the ability to call multiple functions in one request, log probabilities, and reproducible outputs. - Access to world knowledge up until April 2023. - Higher rate limits. According to OpenAI CEO Sam Altman, the price of GPT-4 Turbo is one cent per 1,000 input tokens and three cents per 1,000 output tokens, making it cheaper than GPT-4. Altman also mentioned that cost was prioritized initially, but OpenAI is committed to improving speed in the future. GPT-4 Turbo is currently available for preview to developers with an OpenAI subscription, and OpenAI expects it to be production-ready in the coming weeks. ChatGPT will gradually transition to using GPT-4 Turbo starting from today. At the time of writing, GP-3.5 also received a knowledge update up until January 2022. Another change for ChatGPT users is the removal of the model selector dropdown menu. Instead, ChatGPT will automatically choose which model and modality to use based on user requests, including images and text-to-speech from DALL-E 3. In addition to discussing OpenAI's established goal of achieving artificial general intelligence (AGI), which refers to AI that surpasses human intelligence, Altman emphasized the company's future goal of creating customized AI agents with specific skills. To achieve this, OpenAI is launching GPTs, which are customized versions of ChatGPT. OpenAI announced several new products and features at its annual Developer Conference, including GPT Store, ChatGPT Enterprise, and Assistants API. These offerings are built on OpenAI's powerful generative AI technology and aim to make it easier for developers and users to create and utilize GPT. GPT is a generative application based on OpenAI's GPT-3 model, capable of generating various content such as text, images, music, and code based on user input and preferences. According to Sam Altman, GPT is OpenAI's core product and the future of the company. The GPT Store will be a marketplace where creators can design, customize, and sell their own GPT models. It is set to open later this month, and creators can earn income based on the popularity of their GPT. Canva and Zapier have already gained early access to the GPT builder and have released their own GPT models. With ChatGPT Enterprise, users will be able to create GPTs that can only be shared within their organization. ChatGPT Enterprise will be available on Wednesday, November 8. OpenAI emphasizes that conversations on ChatGPT Enterprise will not be used for model training. According to Gartner's 2023 Market Data Handbook, it is predicted that generative design AI will automate 60% of the design work for new websites and mobile applications by 2026. OpenAI also introduced the Assistants API, a tool for building applications based on generative AI that integrates OpenAI's code interpreter, retrieval, and function calling capabilities. The Assistants API builds upon GPT and provides additional tools. Developers can integrate generative AI into their existing applications and services using the Assistants API. For example, the OpenAI code interpreter can write and generate code, and create files. During the presentation, OpenAI demonstrated a voice-based assistant that can modify a user's OpenAI account in real-time. The Assistants API is currently in the testing phase, and pricing details can be found on the OpenAI website. The OpenAI API is available internationally. OpenAI assures that data and files used with the OpenAI API will not be used for training AI models. Altman believes that the Assistants API is a precursor to AI agents and anticipates that as AI capabilities improve, they will gradually be able to take autonomous actions.