OpenAI unveils new GPT-4 models and custom GPTs bot builder program, among other changes

At its first developer event, 'DevDay',, OpenAI announced a range of developments, and among the most significant, we have the launch of GPT-4 Turbo, an improved version of GPT-4, available for text analysis and text and image comprehension. The pricing varies based on usage, with image processing costing $0.00765 per 1080 x 1080 pixels image, and text-only priced at $0.01/1,000 input tokens and $0.03/1,000 output tokens (this means it will be three times cheaper for developers than current models). The model has a context window of 128K tokens and a knowledge cut-off date of April 2023.

OpenAI will also enable users to create personalized bots, known as "GPTs," even without coding expertise, something similar to what can be done with Poe, which has recently launched a monetization program focused on this. Enterprise customers can develop internal GPTs using OpenAI's knowledge base, and developers can link GPTs to databases or knowledge bases to import external data. The company also plans to launch a GPT Store for users to share their GPT creations, with compensation available for creators of popular GPTs. These creations will be available to ChatGPT Plus and ChatGPT Enterprise customers.

There's also a new Assistants API to help developers build agent-like experiences for various purposes, including retrieving external knowledge or performing specific tasks. OpenAI's text-to-image model, DALL-E 3, is also now accessible via an API with built-in moderation tools, priced at $0.04 per generated image. The company also launched the Audio API, featuring six preset voices and two generative AI models. To protect businesses from copyright claims, OpenAI introduced the Copyright Shield program, covering legal fees for customers facing IP lawsuits related to content created with OpenAI's tools.

Other announcements included assistance in building customer models, the removal of the model picker in ChatGPT for a simpler user experience, and the launch of Whisper large v3, an open-sourced speech recognition model, among several other significant changes that you can explore in full detail in the Keynote video here.