It’s not much of a stretch to say that OpenAI & ChatGPT have taken over the world, and in fact, that’s exactly what its critic are worried about. Supporters and critics alike will want to watch the keynote event from OpenAI’s very first “Dev Day”, where the company made a range of groundbreaking announcements.
The full video is about 45-minutes and well worth watching. It’s more than just the keynote speech of an AI developer event: this is the next great tech company graduating from crawling to walking and these are its first big steps.
Here are all the key announcements from the event.
GPT-4 Turbo is a significant enhancement to previous models, boating a 128K context window and knowledge of global events up to April 2023. Not only is it more advanced, but it also comes at a reduced cost, with input tokens priced at $0.01 per 1K and output tokens at $0.03 per 1K, making it even more accessible to developers.
The upgraded GPT-3.5 Turbo now supports a default 16K context, available at a lower price point. Input tokens cost $0.001 per 1K, and output tokens are $0.002 per 1K. Additionally, fine-tuning is available for this model, enabling more precise and cost-effective AI training.
The new Assistants API is designed to streamline the development of AI-driven applications. It enables the creation of purpose-specific AI assistants capable of complex tasks such as natural language data analysis, code assistance, and more. This API simplifies the integration of various models and tools, enhancing developer productivity.
GPT-4 Turbo now integrates visual input capabilities, allowing for applications such as caption generation and detailed visual analysis. Additionally, DALL·E 3 is available for image generation, and text-to-speech capabilities have been introduced, offering six natural-sounding voices.
The new GPTs feature allows developers to create and customize their own versions of ChatGPT by combining specific instructions, data, and capabilities.
This customization is not only intended to grant developers more control over the AI experience, but pretty soon they’ll be able to make money from their creations. Let the GPT arms race begin!
OpenAI has announced significant price reductions across its platform, making its advanced AI models more affordable for a broader range of users.
Moreover, the increase in rate limits for token usage will enable applications to scale more effectively.
With the introduction of features like reproducible outputs, improved function calling, and JSON mode, developers can now leverage greater consistency and accuracy in their applications.
OpenAI has introduced Copyright Shield, a new initiative to support and defend customers facing copyright infringement claims. This protection is part of OpenAI’s commitment to its users’ legal safety.
There is much more to dig into with the latest OpenAI announcements- this is just skimming the surface. Which of these announcements is most exciting to you? What do you hope comes next?
Read More
OpenAI Dev Day 2023: Keynote Speech
The full video is about 45-minutes and well worth watching. It’s more than just the keynote speech of an AI developer event: this is the next great tech company graduating from crawling to walking and these are its first big steps.
Here are all the key announcements from the event.
(1) Introduction to GPT-4 Turbo
GPT-4 Turbo is a significant enhancement to previous models, boating a 128K context window and knowledge of global events up to April 2023. Not only is it more advanced, but it also comes at a reduced cost, with input tokens priced at $0.01 per 1K and output tokens at $0.03 per 1K, making it even more accessible to developers.
(2) Revamped GPT-3.5 Turbo
The upgraded GPT-3.5 Turbo now supports a default 16K context, available at a lower price point. Input tokens cost $0.001 per 1K, and output tokens are $0.002 per 1K. Additionally, fine-tuning is available for this model, enabling more precise and cost-effective AI training.
(3) Assistants API
The new Assistants API is designed to streamline the development of AI-driven applications. It enables the creation of purpose-specific AI assistants capable of complex tasks such as natural language data analysis, code assistance, and more. This API simplifies the integration of various models and tools, enhancing developer productivity.
(4) Text, Images, and now Audio!
GPT-4 Turbo now integrates visual input capabilities, allowing for applications such as caption generation and detailed visual analysis. Additionally, DALL·E 3 is available for image generation, and text-to-speech capabilities have been introduced, offering six natural-sounding voices.
(5) GPT Store & Customizable GPTs
The new GPTs feature allows developers to create and customize their own versions of ChatGPT by combining specific instructions, data, and capabilities.
This customization is not only intended to grant developers more control over the AI experience, but pretty soon they’ll be able to make money from their creations. Let the GPT arms race begin!
(6) Pricing and Accessibility Enhancements
OpenAI has announced significant price reductions across its platform, making its advanced AI models more affordable for a broader range of users.
Moreover, the increase in rate limits for token usage will enable applications to scale more effectively.
(7) New Developer-Friendly Features
With the introduction of features like reproducible outputs, improved function calling, and JSON mode, developers can now leverage greater consistency and accuracy in their applications.
(8) Copyright Shield
OpenAI has introduced Copyright Shield, a new initiative to support and defend customers facing copyright infringement claims. This protection is part of OpenAI’s commitment to its users’ legal safety.
And even more innovations…
There is much more to dig into with the latest OpenAI announcements- this is just skimming the surface. Which of these announcements is most exciting to you? What do you hope comes next?
Read More
Upvote
0
