Google puts Gemini 4 Argon into its main app, aiming for more direct answers and faster coding workflows

Google is bringing its latest Gemini model, Gemini 4 Argon, into the Gemini app, positioning the update as a practical upgrade for how users get answers and complete tasks. In a product announcement posted Thursday, the company said Argon is designed to provide “frontier reasoning” while improving performance on complex requests, particularly when users need sustained attention across longer work sessions.

Google’s changes are aimed at reducing conversational friction. The company said Argon answers questions more directly with less hedging, is less prone to unnecessary repetition, and avoids what it described as cliché phrasing. For longer tasks, it said Argon provides progress updates more often and more effectively, with the goal of keeping users informed rather than waiting until the end of a multi-step request.

Coding is a second focus. Google said Argon includes major improvements in coding, and it is making it easier to move between chatting and coding by bringing Google Antigravity’s agentic coding interface into the Gemini macOS app. The company described the interface as a way to work through coding tasks within Gemini, while also letting users shift between general conversation and code-focused workflows.

In addition to model and coding changes, Google is updating how users can delegate work. The company said Gemini now includes a simple toggle from chat to “tasks,” and that when users choose tasks, Gemini can work even when they are not at their computer, helping users complete work across Google apps and other services.

Google also emphasized that Gemini’s task execution is increasingly connected to third-party apps. It said Gemini connects users to more than 40 apps, naming services such as Instacart, OpenTable, Adobe, and Shopify. The company’s examples ranged from ordering groceries and booking tables to designing and building business-related materials.

For Alphabet, Gemini remains a central pillar in its broader push to put generative AI into consumer and productivity software. By shipping model improvements directly into the Gemini app and layering in task execution and app integrations, Google is effectively trying to move beyond “chatting” toward AI that can carry out workflows users would otherwise manage manually.

The introduction of an “agentic” coding interface inside the Gemini macOS app underscores that Google is treating coding assistance as a workflow, not just an answer. Agentic coding, in plain terms, refers to an AI-assisted process where the system can help carry out parts of a programming task through a structured interface, rather than only generating code snippets in a single response.

Still, the announcement leaves several questions unanswered that can matter to users evaluating whether the update fits their needs. Google did not specify how Argon’s rollout will expand beyond the Gemini app or what the availability timing looks like across devices and regions. It also did not provide details on any differences between the experience on macOS versus other platforms, nor did it disclose limits, reliability metrics, or cost changes associated with Argon beyond the mention of AI Pro and Ultra subscriber access.

Going forward, the key thing to watch is whether Google’s described improvements translate into measurable changes in day-to-day productivity, especially on longer, multi-step requests and coding tasks where users tend to notice repetition, stalled progress, or unclear next steps. Users and developers will also be watching whether Gemini’s app connections to more than 40 services continue to expand and how the “tasks” toggle behaves when users are offline.

keyFacts