On May 13, OpenAI unveiled its latest flagship AI model: GPT-4o, marking a significant milestone in the industry of AI!
The abbreviation "o" in GPT-4o stands for "Omni", indicating an all-capable system. Although the GPT-5 or the AI+ search, which had been long anticipated, was not the main focus of this launch event, GPT-4o represents a significant advancement in human-machine interaction.
Building upon GPT-4, OpenAI has bolstered the AI's low-latency processing capabilities for text, audio, and visual inputs. This enhancement enables GPT-4 to comprehend and respond to human needs more effectively through multimodal perception.
Real-Time Interactive
During the live demonstration at GPT-4o's launch event, the model showcased its impressive capability to respond to audio inputs within 232 milliseconds.
Unlike its predecessors, which would abruptly terminate conversations due to brief pauses, GPT-4o exhibited an understanding of common human conversational patterns, such as pauses and moments of thought, allowing for more natural and seamless interactions.
Real-Time Vision
At a recent event, Brockman, an engineer from OpenAI, conducted a demonstration involving AI models engaging in conversation.
He placed an older version of GPT, which lacks visual capabilities, alongside a newer version with camera support, allowing them to interact verbally. Brockman enabled the camera on the newer model, observing from the sidelines.
Leveraging its visual abilities, the camera-enabled GPT model accurately identified a light bulb behind the presenter. Remarkably, it also exhibited self-awareness, recognizing its conversation partner as an older GPT iteration. The two models engaged in a natural, friendly exchange.
While the two AI were conversing, a lady discreetly made the ✌🏻 sign behind Brockman's back.
Then Brockman asked GPT-4o whether any events had occurred in this room. The AI model, without hesitation, mentioned the presence of a second individual and described their actions.
Remarkable, indeed remarkable, the integration of Artificial Intelligence (AI) into security systems has transitioned from a conceptual idea to a tangible reality.
Real-Time Study Assistant
It can not only provide answers to simple queries about human actions but also handle more complex mathematical and graphical problems.
This capability should be a boon for parents seeking assistance with their children's studies. GPT-4 can essentially take on the role of a home tutor, solving mathematical problems on demand! What's truly remarkable is that this was a real-time, live demonstration for hundreds of people, not a post-edited showcase (yes, we're looking at you, Google Gemini).
It's safe to say that OpenAI still has many more powerful capabilities up its sleeve that have yet to be revealed.
Benchmark Test
The real-time demonstration has already been remarkably impressive, and based on the benchmark scores on paper, GPT-4o has consistently outperformed others.
In text evaluation tests, GPT-4o significantly outperformed numerous models, including Claude 3 Opus, Google Gemini Pro 1.5, and even Meta's "open-source version of GPT-4," Llama 3 400B.
In the evaluation of the model's visual understanding capabilities, GPT-4o has demonstrated superior performance and outperformed competitors by a significant margin across various metrics.
Once again, OpenAI has solidified its position as a dominant force in the field of artificial intelligence.
No Additional Charges for Enhanced Features
At nearly every product launch event, OpenAI has announced price reductions, and this occasion is no exception.
GPT-4o and all services previously included in the paid ChatGPT Plus subscription, encompassing visual capabilities, internet access, GPT Store, and other functionalities, will be made available to all 100+ million registered ChatGPT users in the near future, at no additional cost.
OpenAI humorously remarked that charging for GPT-4 was never their intention, implying a desire to provide their services freely, albeit limited by resource constraints.
Regarding the pricing of APIs, the company has reduced the costs of their models that leverage GPT technology.
At the moment, free members do not have access to experience the GPT-4o model.
How to Use GPT-4o With Unlimited Right Now?
I understand your eagerness, so let me provide you with a method to try the latest GPT-4o for free!
Click the URL and switch the model on the bottom left.
To initiate a conversation, simply type your message into the text box. Additionally, you can click on the image button located on the left side of the chat window to upload images for AI recognition.
This feature will allow you to engage in a dialogue centered around the content of the uploaded images.
No need to pay $20 per month, every user of Anakin.ai will gain 30 free credits per day and chat with GPT-4o! No conversation limits!