Chat GPT-4o Version: Voice, Image, And More…
Enhance Your Conversations Now with Chat GPT-4o Version
OpenAI’s ChatGPT is now based on the new Chat GPT-4o language model. The letter “o” at the end of the name means omni (omnichannel), which emphasizes the versatility of the model. It is capable of understanding speech and music (that is, sounds), generating speech and music, as well as images and text.
The new Chat GPT-4o Version offers a range of innovative features that elevate the conversational AI experience to a new level. One of the most notable additions is the Voice feature, which allows users to interact with the AI through spoken commands and responses. This not only streamlines the user experience but also adds a new dimension of interactivity to the conversations. Additionally, the Image feature enables users to communicate with the AI using visual inputs, such as photos or screenshots. This feature opens up a whole new realm of possibilities for users to convey information and receive tailored responses from the AI.
Furthermore, the Chat GPT-4o Version also boasts a range of other enhancements that make it a truly versatile and efficient conversational AI tool. The incorporation of advanced natural language processing algorithms ensures that the AI is able to understand and respond to a wide variety of queries and prompts in a nuanced and context-aware manner. Additionally, the AI has been trained on a vast repository of data, which allows it to provide accurate and relevant information on a wide range of topics. This comprehensive knowledge base makes the AI a valuable resource for users seeking quick and reliable information on various subjects.
Overall, the Chat GPT-4o Version represents a significant advancement in the field of conversational AI, offering users a more intuitive and interactive experience. With features like Voice, Image, and advanced natural language processing capabilities, the AI is able to engage with users in a more natural and seamless manner. Whether users are seeking information, assistance, or simply engaging in casual conversation, the Chat GPT-4o Version is a powerful tool that can enhance communication and productivity.
Judging by the tests, GPT-4o is superior to its other versions and competitors in 4 out of 6 parameters.
But… there is also an important gap between GPT-4o and its competitors. The new model can work with a context window up to 128 thousand tokens (this corresponds to 300 pages of text). That is, it has not been increased compared to GPT-4 Turbo (you will find information about this version of the language model below). At the same time, Claude 2.1 has a context window of 200 thousand tokens, and Gemini 1.5 has 1 million tokens.
The new version will be available to everyone for free, but don’t rush – so far the capacity and test limits are only enough for paid users of the system: the capabilities of GPT-4o “will be rolled out gradually, starting today.”
When a new version of ChatGPT is released with an improved GPT-4o model, in the first days after the release, there is a high probability that free users will not be able to access it. This is due to expected increased demand. At the same time, the company warned in advance that if the load is high, free users will be automatically switched back to the previous version of GPT-3.5.
Therefore, most likely, in the first days and even weeks after the release of GPT-4o, access to the new model will mainly be given to paid subscribers who do not have such restrictions. Free users will either have to be patient or consider paying for a subscription to try out ChatGPT’s improved features as quickly as possible.
General Features of Chat GPT-4o Version
When the new version of GPT-4o becomes publicly available, all users will get:
- Response to voice input is 320 milliseconds on average and 232 milliseconds at peak; this is comparable to the human response.
- During vocal communication, there is a change in intonation.
- The neural network remembers all conversations with the user and you can make references to different moments of communication.
- Supports 50 languages.
- It can synthesize objects in 3D.
- Improved analysis of images, including graphs, charts and screenshots.
- Actively uses its own “knowledge” and information from open sources (Internet).
Before GPT-4o, the average latency of ChatGPT voice communication was 2.8 seconds (GPT-3.5) and 5.4 seconds (GPT-4). This is because in these two models the audio mode is a pipeline of three separate models: a simple model converts the audio to text, GPT-3.5 or GPT-4 ingests and outputs the text, and a third simple model recycles that text. to the voice. Therefore, the neural network loses some information: In one version of OpenAI, it cannot directly read the tone of speech, understand multiple speakers, or recognize background noise. “All data is now processed by a single model.”
Let’s remember that the previous OpenAI language model was called GPT-4 Turbo and appeared not so long ago, in September 2023.
ChatGPT update in September 2023
On September 25, OpenAI published an article on their blog in which they talked about updating their neural network service. In this update we do not see an improvement in the neural network algorithms themselves, but there is something equally important in it – the first steps towards the transition of ChatGPT to a multimodal model.
Multimodal models are algorithms that can work not only with text, but also with pictures, video and audio. In the article about the GPT-4 language model, we wrote that according to forecasts, it should already become multimodal. And now the predictions came true – the update endowed the neural network with the ability to synthesize and recognize speech, as well as image analysis.
It was reported that voice input and audio generation are available only to users of mobile versions of ChatGPT for iOS and Android, but work with pictures has appeared on all platforms.
How to use these features?
Voice input is available in the neural network mobile application, but this will not surprise anyone. Speech synthesis is not so easy to enable. To do this, go to the mobile application, go to Settings, and then click New Features and select Voice conversations. Then press the headphone button located in the upper right corner of the main screen and select one of the five voices that the neural network will speak to.
To use the image feature, OpenAI says, you need to click the photo icon to the left of the query input panel. In the mobile application, you will first need to click the “+” icon.
