Google has released an early version of Gemini for Mac for closed testing. A separate macOS app should give the service faster access to Apple users, where ChatGPT and Claude already have their own clients.
Google has released an early version of Gemini for Mac for closed testing. A separate macOS app should give the service faster access to Apple users, where ChatGPT and Claude already have their own clients.
According to Bloomberg, Google this week began distributing an early build of Gemini for Mac to participants in its consumer beta program. In a message to testers, the company explicitly stated that this is not yet the final version and contains only critical features, not the full set of features that are planned to be added later.
Google needs to launch a separate client for macOS not only for convenience. OpenAI and Anthropic already have their own ChatGPT and Claude apps for Mac, so Gemini is actually catching up with competitors in the fight for the user who wants to access AI without a browser and quickly upload documents in their usual workflow.
In the test version, Google is testing the generation of images, videos, music, tables and graphs, mathematical problems and information analysis. The application is also able to search for data on the Internet, work with previous dialogues, personalize responses and analyze downloaded files. One of the most interesting features should be Desktop Intelligence. It allows Gemini to get context from other applications on the Mac, for example, from the calendar, and use this data for more personalized responses.
In the app code that the researchers examined, Google explains this feature as allowing Gemini to see what the user sees, including the context of the screen, and pull content from authorized apps only when the service is running. This brings Gemini closer to scenarios that some competitors already offer on macOS.
Bloomberg writes that the internal name of the project is Janus. The start of external testing usually means that the product is already being prepared for release, although Google has not yet announced an exact launch date for all users.
Previously, dev.ua wrote about how Google introduced Gemini 3.1 Flash-Lite — a model that it positions as the fastest and most economical in the Gemini 3 series for long-running scenarios such as translation, moderation, and other tasks with a large number of requests.
