Google says the model is now available in the app to Google AI Pro and Ultra subscribers. The announcement also describes a toggle from chat to tasks, under which Gemini can work while a user is away from the computer. Google says Gemini connects to more than 40 apps, including Instacart, OpenTable, Adobe, and Shopify.
The mechanism: Google places model access, task execution, and connections to more than 40 apps in one app experience. This shifts the point of use from model access alone toward a bundled interface for delegating work across connected services. Why now: The update marks a move beyond the earlier restricted cyber-defender availability of Argon and makes it relevant to paying consumer subscribers. The announcement describes the intended product surface, not evidence that tasks complete reliably or successfully.
The first implication: For builders, the distribution and integration surface may matter as much as the model’s reasoning claims when users delegate work. The evidence boundary: Google’s announcement describes product capabilities but gives no independent performance results, detailed regional coverage, or app-by-app eligibility and task limits. The first watch condition: Whether Google publishes concrete task-completion examples, availability details, or safeguards for actions across connected apps. This is a distinct analytical record because no prior New Runtime analysis was supplied for this specific move from restricted access to the Gemini app.