During the Spring Update event, OpenAI unveiled its latest flagship model, GPT-4o (‘Omni’), which will be accessible to all users, including both free and paid ChatGPT users. This means that free users now have access to GPT-4 class intelligence at no cost. Additionally, all tools and premium features of ChatGPT Plus will be available to free users.
Some of the features now available to free users include accessing the internet on ChatGPT, uploading images and utilizing the vision capabilities of GPT-4o, uploading and analyzing files and documents, creating charts, performing Advanced Data Analysis (previously known as Code Interpreter), enabling the Memory feature, and accessing GPTs and the GPT Store.
Essentially, OpenAI is bringing all paid features to the free version of ChatGPT. The new model will be rolled out to all users in the coming weeks. However, there is a limit on the number of messages for free users. Once this limit is reached, users will be switched to the GPT-3.5 model automatically.
GPT-4o is Truly Multimodal
One of the most notable aspects of GPT-4o is that it is a multimodal model designed from the ground up. Previously, OpenAI used different models for different modalities, leading to increased latency and a subpar user experience. For example, during Voice Chat, OpenAI used Whisper for voice processing, GPT-4V for vision, and GPT-4 for text processing and reasoning.
However, the GPT-4o model can simultaneously process all three modalities – text, audio, and vision and intelligently reason across them. In some of the demos showcased during the event, the experience was likened to scenes from the movie Her. It’s worth noting that a good internet connection is necessary to fully experience this capability.
The GPT-4o model has the ability to perceive visual information in real-time and express emotions in a natural manner, with a range of tones. This makes conversations feel less robotic and more spontaneous. Moreover, you can now start speaking to interrupt the model and seamlessly continue the conversation.
Furthermore, the GPT-4o model can understand the emotions conveyed through your voice. For instance, if you sound anxious and are breathing rapidly, it may suggest calming down. Additionally, the model can translate languages in real-time. OpenAI has stated that the new GPT-4o model supports 50 languages.
ChatGPT Introduces Desktop Application for macOS
ChatGPT has launched a desktop application for macOS, allowing users to voice chat with the AI on their Mac computers. The macOS app now includes a vision capability, enabling it to see the user’s screen. This feature is particularly useful for coding, as ChatGPT can now view and reason with code. It is uncertain whether a similar ChatGPT app will be released for Windows.
For developers, the new GPT-4o model will be accessible through the API. It offers a 50% lower cost, 2x performance improvement, and a 5x higher rate limit compared to GPT-4 Turbo.
With OpenAI making all paid features available to free users, you might wonder what remains for ChatGPT Plus subscribers. Paid users will still have 5x the capacity of free users. Additionally, Mira Murati announced at the end of the event that OpenAI will soon release the next “frontier” model. This means that paid users will have early access to the upcoming advancements in AI technology.
0 Comments