OpenAI's Sam Altman at their DevDay event shared updates about their latest version GPT-4 Turbo.
Here, let me break down the improvements and benefits of GPT-4 Turbo briefly.
1. Upgraded Knowledge Base and Longer Context Window
GPT-4 Turbo extends its knowledge base to April 2023. This is a leap forward from its predecessor, which stopped in September 2021.
A longer context window, now at 128k tokens, empowers this model to understand and provide more detailed responses.
Users can ask complex questions, and even instruct the model. It is to use specific coding languages for tailored results.
2. Price Reduction for Affordability
Developers will enjoy substantial cost savings with GPT-4 Turbo. Input tokens are now three times cheaper at $0.01 per 1,000 tokens.
Output tokens have also become more affordable, reducing to $0.03 per 1,000 tokens.
This price reduction aims to make advanced AI models more accessible to developers.
3. Image Prompts and Text-to-Speech
GPT-4 Turbo introduces image prompts. This will allow users to input images for generating captions or descriptions.
It also handles text-to-speech requests, with a variety of preset voices for users to choose from.
This expanded functionality opens new possibilities for creative interactions with the model.
4. Custom Chatbots and GPT Store
OpenAI now lets users create custom chatbots using their platform. This feature was previously limited to enterprise users.
The GPT Store will be OpenAI's new app store. It will allow users to create and share personalized AI chatbot models.
Users may even earn money based on the usage of their creations, and innovation in AI development.
5. Copyright Shield: Protecting Users
OpenAI is following the steps of tech giants like Microsoft and Google by introducing Copyright Shield.
This program provides legal protection to users facing copyright infringement claims related to their AI applications.
OpenAI will defend customers and cover incurred legal costs. This will ensure a relevently safe environment for AI innovation.
6. Access and Rate Limits
Access to GPT-4 Turbo is available to 'all paying developers' and can be accessed through the OpenAI API.
Rate limits for GPT models via the OpenAI API have been clarified, with GPT-4 Turbo currently having rate limits of 20 requests per minute and 100 requests per day during the preview stage.
A production-ready version is set to be released soon.
Finally, as OpenAI continues to evolve, the future of AI looks more promising than ever.
👉 Watch the Full YouTube Video: OpenAI DevDay, Opening Keynote