In late 2023, Open AI announced many vital improvements and changes to its platform and introduced GPT-4 Turbo.

GPT-4 Turbo is a new version of Chat GPT that is very fast, cheaper, and includes vital benefits. The announced novelties make it possible to conclude the direction of the development of Chat GPT in 2024. Below, we’d like to describe the most important things.

1. Larger context size.

GPT-4 Turbo will support a large context window size of 128,000 tokens (words, spaces, sentences) used in all queries until the model “forgets” the entire conversation. This is equivalent to about 300 pages or one standard book. GPT-4 Turbo has increased response accuracy due to the greater context provided.

2. Extract information from documents.

The new version of GPT will allow users to upload documents and databases to feed the model with various information to expand its access to up-to-date and specific knowledge. We can now upload our contracts and invoices and discuss them with Chat GPT.

On the other hand, Open AI updates the available information, and the chatbot knows the world as of April 2023.

It is important to note here that Grok – the new artificial intelligence of X (the former social platform Twitter) – works with up-to-date information from tweets up to the last second and thus probably has a competitive advantage.

3. GPT now goes visual and has audio.

GPT-4 Turbo can read images and understand precisely what is in the pictures. On the other hand, generating images through DALL·E 3 will be possible for developers using the Chat GPT API.

Companies like Snap, Coca-Cola, and Shutterstock use DALL·E 3 to generate images and designs for their clients and campaigns programmatically. Open AI offers a variety of format and quality options, with prices starting at $0.04 per generated image.

Text-to-speech (TTS) will also be supported, creating audio imitating a human voice from typed text. Open AI also released its new version of the voice recognition and text conversion model, Whisper V3, which continues to be open source.


    4. Improved user control and protection.

    Chat GPT-4 will offer a functionality called reproducible output. This makes it possible to give specific answers to certain basic prompts (user questions).

    Protection in disputes over copyright abuse is also introduced. Open AI will cover its customers’ costs if there are lawsuits and fines against them for infringement of rights when using Chat GPT to create content. Chat GPT Enterprise will also need to train its language models on data from its enterprise customers.

    5. Improvements for software engineers.

    Open AI’s strategy is to make the development of applications on its platform and programming interfaces (APIs) extremely easy and convenient. Improved function calling and calling multiple functions simultaneously are some of the benefits available to developers. The limit of tokens used by engineers developing applications with the Chat GPT API has also been increased. You can also request an additional volume of tokens if you need to.

    We’d like to remind you that tokens are the unit of measure for using the platform – one token corresponds to a syllable or part of a word that the language model generates. We want t.

    6. Custom models.

    It is now possible to train models with completely new domains of application and knowledge. Large companies (together with Open AI experts) will be able to train their models with large datasets and train them to perform specific tasks in the companies’ fields of activity.
    At this stage, it is a long and expensive process that only top companies with leading market positions can afford. Experimental programs for the so-called are already working for them. finetuning (specific training) of the models.

    7. Decreasing the price of tokens.

    The cost of GPT-4 Turbo (and GPT -3.5 Turbo) drops significantly. One thousand tokens generated by the advanced models will be priced at $0.03, and the base tokens will be priced at $0.006. This will make Chat GPT much more accessible to developers, enterprise customers, and ordinary users.

    8. Improved interface and functionality.

    There will no longer be a need to choose different models and tools (plugins) for individual tasks. Chat GPT will be able to understand what you want to do and activate the most appropriate model to perform the required functionalities. Depending on your questions, the chatbot will search the Internet for information, analyze data, create images, write software code, etc.

    This is related to the direction of the development of artificial intelligence in the coming years. Before long, we can tell computers what we want from them, and they will perform the corresponding tasks.

    9. GPTs are introduced.

    GPTs are the new name for models or agents that are specialized and trained to solve tasks and create value in distinct domains. This will allow users and developers to make a customized version of Chat GPT for almost anything. We will be able to give specific instructions for the operation of the AI ​​agent, feed it with particular data, and formulate scenarios of functionalities and user actions.
    Plugins go to history and become GPTs to perform individual tasks.
    Logo Creator

    10. GPT Builder – we write applications using natural (human) language.

    Digital content creation (logo, application, presentation, etc.) will occur after entering a description and instructions in natural language.

    Imagine two parts of the screen – in one, you describe how your digital project should look and work, and in the other, a visual representation of the expected result is created in real time. Adding application configuration options, custom actions and functionalities, and more is possible.

    Improve your digital and business skills with our on-demand courses


    11. GPTs Store.

    This is Open AI’s App Store. Developers can create custom AI agents and models (GPTs) and upload them to the GPTs Store platform. The creator companies will share their revenue with Open AI, similar to the commissions that Apple takes for payments in installed applications on the iPhone (In-app purchase).

    See a commentary video from the event to introduce the novelties OpenAof I DevDay announcements here:

    Author: Emiliyan Enev

    Емилиян Енев

    Emilian Enev is the chairman of the Startup Factory association and an entrepreneur in the field of new technologies.

    He is one of the founders of the blockchain company Recheck, which deals with adapting the technology to the needs of different businesses.

    Emilian has been involved in the development of several online projects and digital products, with his strengths in business management, online marketing, content creation, idea validation, monetization strategies and customer engagement.

    Do you want to start a career in IT and Software Quality Assurance?

    Prepare yourself with our latest on-demand video course. Check out what you can expect from the course ⬇️

    Лого БФЖ

    Startup Factory is a non-profit organization founded in March 2015 by a group of young entrepreneurs working in the IT sector. Startup Factory helps people develop their digital and business skills. Together, we build a supportive community.  Check out our platform with on-demand video courses and valuable free resources.


    Sign up for our monthly newsletter and be the first to know about events, trainings and opportunities in the digital world.