Discover the exciting developments and upgrades in GPT-4, as unveiled recently by Sam Altman at the OpenAI Dev Day. OpenAI‘s latest innovations promise to transform the landscape of AI models, offering enhanced capabilities and functionalities for developers and AI enthusiasts. This comprehensive list outlines the major highlights and advancements that were discussed at the event, providing you with a sneak peek into the future of AI technology.
- GPT-4 Turbo: This new model offers several improvements, including a significantly larger context window of up to 128,000 tokens, improved function calling, and the introduction of “JSON mode” for easier API use.
- Better Knowledge: GPT-4 Turbo offers knowledge with a cutoff window extended through April 2023 and allows users to inject their own information from external documents or databases into the model.
- New Modalities: GPT-4 Turbo integrates DALL·E 3, GPT-4 with vision capabilities, and a new text-to-speech model, enabling the use of images as inputs and generating captions, classifications, and natural-sounding audio.
- Customization: Users can work with OpenAI to fine-tune models specific to their business needs.
- Higher Rate Limits: Rate limits are increased, making it easier for established GPT-4 customers to perform more tasks.
- Copyright Shield: OpenAI offers protection against potential legal issues related to copyright for apps using their API, providing reassurance for developers.
- Pricing: GPT-4 Turbo is considerably cheaper than GPT-3, with pricing set at 1 cent per 1,000 prompt tokens and 3 cents per 1,000 completion tokens.
- GPTs and GPT Store: OpenAI is introducing GPTs, customizable AI assistants, and an associated GPT Store where users can create, share, and potentially sell their GPTs.
- Assistance API: Users can now create custom AI assistants within the API with features such as persistent threads for context retention.
These upgrades and features demonstrate OpenAI’s commitment to enhancing AI models, offering more control to developers, and expanding the possibilities for AI applications.