Table of Contents
One of the most interesting new features is the expansion to 128,000 context windows in the GPT-4 Turbo compared to 8,000 in the GPT-4.
For GPT-4 Turbo users, this means they can process 16 times as much text at once, equivalent to 300 pages of text – enough space to include small novels like Animal Farm and The Great Gatsby.
In his keynote, Sam Altman, CEO of OpenAI, also announced the end of GPT-4 ‘s knowledge limitation. While GPT-4 was able to access information about events through 2021, the AI giant announced that GPT-4 Turbo will know the world until April 2023.
This comes shortly after OpenAI announced that ChatGPT had reached 100 million weekly active users.
Taken together, these recent announcements underscore that GPT-4 Turbo not only has access to an enhanced knowledge base, but also has the power to respond to larger user requests.
- OpenAI introduces GPT-4 Turbo, an improved version of GPT-4 with a 128,000 context window that is 16 times larger than its predecessor and enables richer text processing.
- GPT-4 Turbo has expanded its knowledge base to include information through April 2023, making its worldview more current.
- New features in GPT-4 Turbo include text and photo analysis, image creation with DALL-E 3, and improved support for function calls and user instructions.
- OpenAI introduces the GPT Store, where you can create custom versions of ChatGPT and monetize them. This opens the door to a variety of potential use cases and makes 2024 an exciting year for AI development.
What does GPT-4 Turbo mean for generative AI?
The improvements OpenAI has made to GPT-4 Turbo make it arguably the best multimodal LLM on the market. With the latest version, GPT-4 Turbo can analyze text and photos as well as create images with DALL-E 3.
This means it can be used for a variety of use cases, from answering user questions to creating written content and images, analyzing text/photo, creating captions and much more.
However, one area where GPT-4 Turbo is particularly sophisticated is its support for function calls and the ability to follow user instructions.
This means you can trigger multiple actions with a single prompt, e.g. B. “ Open the car window and turn off the air conditioning , ” which previously had to be split into several commands. It is also better at executing instructions and responding in certain formats such as XML.
Considering that GPT-4 was already at the top of AI performance benchmarks such as Multi-Task Language Understanding (MMLU), HellaSwag, and Common Sense Reasoning on ARC, the improvements provided by GPT-4 Turbo have the potential to improve OpenAI ‘s position to solidify itself as a leading force in the generative AI market.
This is especially true considering the low running costs of GPT-4 Turbo. According to OpenAI, GPT-4 Turbo requires three times fewer input tokens (0.01 USD) and two times fewer output tokens (0.03 USD) than GPT-4.
GPTs from OpenAI: better customization
GPTs (Generative Pre-Trained Transformers) from OpenAI also have the potential to develop into a strong driving force for the company.
GPTs allow one to create a custom version of ChatGPT that can be connected to a number of external data sources and used to perform specific tasks.
OpenAI said in its official announcement:
“ Anyone can easily create their own GPT – without programming. You can create it for yourself, for internal use within your company only, or for anyone. It’s as simple as starting a conversation, giving him instructions and additional knowledge, and specifying what he can do, e.g. B. to search the Internet, generate images or analyze data. “
In a move that some commentators have likened to the opening of the App Store for iPhone, turning a smartphone with a few apps into an all-in-one tool for much of our digital lives, the GPT Store is coming soon allows anyone to upload and monetize custom GPTs.
The company revealed:
“ Later this month we are launching the GPT Store, which will feature creations from verified creators. Once GPTs are in the store, they become searchable and can move up the leaderboards. We will also feature the most useful and beautiful GPTs in categories such as productivity, education, and “ just for fun . ” You will also be able to make money in the coming months depending on how many people use your GPT. “
So far, ChatGPT Plus and Enterprise users have two sample GPTs available to experiment with. The first, from Canva, allows users to create visual designs using natural language prompts.
The second, Zapier’s AI Actions, allows developers to create a personalized GPT in ChatGPT that can be connected to over 6,000 Zapier apps.
When building custom GPTs, developers can use a code interpreter and retrieval tool using the Wizard API.
Code Interpreter allows programmers to write and run Python code in a sandbox execution environment. This allows processing files and generating graphics, charts and code that can solve complex code and mathematical problems.
The retrieval tool can collect data from external sources, e.g. B. from proprietary data sets to gain insights from third-party systems.
Ultimately, GPTs will make it easier for developers and companies to create their application-specific versions of ChatGPT that can streamline specific processes and workflows.
With GPT-4 Turbo, ChatGPT solidifies its position as the leading generative AI assistant on the market.
The extended context length, along with the multimodal capabilities and customization capabilities of GPTs, has opened the door to a wider range of potential use cases than ever before.
With a crowd-sourced approach to improving GPTs of all kinds, 2024 will be a turbo year for AI.