Microsoft Copilot was further improved this morning as it is now powered by the latest GPT-4 Turbo model for text processing and the latest DALL-E 3 model for text to image processing. Copilot with GPT-4 Turbo is being tested by a small number of users and will be available to everyone in just a few weeks. This capability can be accessed by Windows 11 users directly or other users through the Copilot tab in Microsoft Edge.
As we previously reported, GPT-4 Turbo provides improved processing of user-entered text as it has the ability to process 128,000 simultaneous context tokens which is equivalent to a 300-page prom. It has also been trained with more recent data up to April 2023. Users can also upload their own data to be processed by this generative AI.
Copilot can also do a summary for the video being watched. In the example given, Copilot can tell a summary of the currently playing video complete with a timestamp so that the user can skip to the summarized part of the video. Through the multi-modal capabilities of GPT-4 Turbo, integration with Bing Image is also done to provide more accurate search capabilities using images.
The latest DALL-E 3 model can be accessed through Copilot and also Bing Image Creator. With more recent models, the images generated by DALL-E will look more realistic due to better context understanding. The difference between images generated using DALL-E 3 and the latest model of DALL-E 3 is as below.
With the use of GPT-4 Turbo, the Code Interpreter feature is also provided with it being able to produce better code based on prom, perform more accurate calculations, mathematical understanding and analyze data. At the moment it is still in the testing phase with it being given to a small number of users to gather initial feedback.