15 мая 2024

Discover the Power of GPT-4o: New Features and Emotion Recognition

OpenAI launched a new version of ChatGPT named GPT-4o on May 13th, 2024, and it's generating a lot of excitement for its potential to revolutionize human-computer interaction. Let’s take a look at its new possibilities, features and how GPT-4o can recognize emotions.

About GPT-4o

GPT-4o (“o” for “omni”) is a new AI model designed for smoother interaction between humans and computers. It can take in information through text, audio, images, and even videos, and respond in the same formats. Interestingly, its response speed is on par with humans, taking around 232 to 320 milliseconds on average.

While similar to previous models in its ability to handle English text and code, GPT-4o excels in understanding non-English languages and is significantly faster and cheaper to run. It also boasts a major advantage in processing visual and audio data compared to existing AI models.

In simpler terms, GPT-4o is like a supercharged AI assistant that can understand and respond to you in a more natural way, using different communication styles and languages.

GPT-4o New Possibilities

  • Real-time conversation. GPT-4o allows for natural back-and-forth conversations across different mediums like text, voice, and even images. GPT-4o can be a virtual assistant that can understand your tone and respond accordingly, or a chat system that feels more natural and engaging.

  • Enhanced accessibility. With its multimodal capabilities, GPT-4o has the potential to improve accessibility for people with disabilities. For instance, it could provide real-time audio descriptions of visual content or translate sign language into spoken language. You can learn more from this video Be My Eyes Accessibility with GPT-4o

  • Boosted creativity. Imagine brainstorming with an AI that can not only understand your ideas but also generate creative text formats, images, or even sounds to complement them. This could be a game-changer for artists, writers, and designers.

GPT-4o New Features

  • Multimodal generation. This is GPT-4o's party trick. It can combine different formats in its outputs. Need a presentation with text, images, and even a voiceover? GPT-4o could potentially generate it all.

  • Improved multilingual support. GPT-4o understands and creates content in over 50 languages, making it a valuable tool for communication across borders.

  • Integrated vision. It can analyze and understand what it "sees" through a camera or computer screen. This opens doors for applications like real-time scene descriptions for visually impaired users or AI co-pilots that can assist with tasks.

GPT-4o Can Recognize Emotions

GPT-4o's emotional recognition capabilities are a result of its innovative design. Here's a deeper dive into how it works:

  • Multimodal inputs. Unlike previous models that process text, audio, and vision separately, GPT-4o handles them all simultaneously. This allows it to analyze how these elements work together to convey emotions.

  • Text analysis. GPT-4o can analyze the sentiment of written text. It can identify keywords associated with emotions (e.g., "happy," "sad"), recognize exclamation points and emojis as emotional cues, and even understand sarcasm through context.

  • Audio cues. It can pick up on vocal variations that indicate emotion. For instance, GPT-4o might recognize a raised voice as anger or a trembling tone as fear.

  • Facial recognition. Through visual input, GPT-4o can detect facial expressions often linked with emotions. A furrowed brow might suggest concentration or confusion, while a wide smile could indicate happiness.

It's important to remember GPT-4o is still under development. While it can process these emotional signals, it has limitations:

  • Accuracy. Interpreting emotions can be complex, especially with subtleties. GPT-4o might struggle with sarcasm or misunderstand nonverbal cues.

  • Context Dependence. Just like humans, GPT-4o relies on context to understand emotions. A raised voice could be anger or excitement depending on the situation.

Emotion Recognition Technology

As experts in facial recognition technology, we are intrigued by the emotion recognition feature in GPT-4o. We are particularly interested in the accuracy of this feature, its potential future applications, and other details that OpenAI has not yet disclosed. We will continue to monitor the development of this feature.

If you are a developer interested in creating your own emotion recognition application or integrating this technology into your business, we recommend reading these articles:


GPT-4o represents a significant step forward in AI's ability to recognize emotions. While not perfect, it paves the way for future AI with a more nuanced understanding of human communication.