ChatGPT for macOS was recently updated with a couple of new UI enhancements. First, Projects, previously announced during the 12 Days of OpenAI, are now supported on Mac as well. However, in this version, you can only browse projects, create new chats, view custom instructions, or access files. You cannot create new projects, modify instructions, or upload new files yet. Additionally, model selection is not available at this time, so it defaults to 4o.
In the settings, there’s a new option to select the map provider, allowing you to switch between Apple Maps or Google Maps. However, when you search for certain location-related queries, the map is displayed through built-in maps powered by Foursquare. When you click on a specific location and request directions, you’ll be redirected to the default map provider you’ve configured in the settings.
Lastly, the advanced voice mode with vision seems to be functioning, albeit still under a feature flag and not publicly available. When enabled, it’s possible to share your screen, and ChatGPT will be able to see the content, similar to how it’s done on mobile devices. However, the current implementation is somewhat slow, and it occasionally doesn’t work as expected. It’s still unclear whether this is due to additional development being required or if it’s just temporary. We’ll have to wait and see if this feature will be announced during the remaining two days of OpenAI’s events. Despite there being many other possible announcements, it’s a clear indication that advanced voice mode with vision for the desktop is close to being released.