The newly released GPT-4o by OpenAI boasts incredible speed, raising questions about whether ChatGPT remains the king of language models.
When Google launched Gemini, it was touted as 5 times more powerful and much faster than GPT-4, while using fewer GPUs. In response, OpenAI introduced GPT-4o, significantly faster than GPT-4, making its predecessor seem outdated.
Performance and Capabilities
GPT-4o can generate 488-word responses in under 12 seconds, a task that takes nearly a minute with GPT-4. This speed improvement is noteworthy, with GPT-4o performing twice as fast as GPT-4 Turbo. It can create a CSV file with information on the world’s 50 largest cities in under a minute, a task that GPT-4 struggles with. OpenAI’s CTO, Mira Murati, highlighted improvements in text, video, and audio processing.
Features and Interactivity
GPT-4o offers real-time bilingual translation and enhanced interaction through natural conversation on smartphones and PCs. It includes voice assistants that compete with Amazon’s Alexa, capable of mimicking human traits like interrupting and understanding tone. In a demo, GPT-4o humorously responded to a user’s heavy breathing, showcasing its ability to comprehend and react to human nuances.
Emotion and Context Understanding
GPT-4o can understand and express emotions, such as interpreting a selfie’s context or responding to a user’s smile during a demo. It also allows users to interrupt and redirect its responses, providing real-time, context-aware answers.
Practical Applications and Accessibility
GPT-4o supports 50 languages, covering 97% of the global population, and is available through an API, allowing developers to build custom models at half the cost and double the speed. This update marks a significant advancement, potentially outperforming Siri, Google Assistant, and Alexa.
As AI development accelerates, with Apple rumored to be working on Siri 2.0 and Google I/O imminent, the competition in AI technology is intensifying.