Google brings Gemini 4 Argon to Pro and Ultra subscribers with tasks mode

SiliconANGLE reported that Argon changes how Gemini answers in the app. Google said responses are more direct and less hedged, and the model repeats itself only when needed. Longer jobs should produce progress updates more often. Tasks mode is switched on with a toggle in the main chat window; once enabled, Google said Gemini can work even when the user is not at the computer. More than 40 connected apps are available, allowing tasks to run as far as a grocery order through Instacart or a restaurant booking on OpenTable. Adobe and Shopify are on the list for people designing or running a business.

The model also brings what Google called major improvements in coding. At its Sept. 30 unveiling, Google showed Argon at 77.9% on DeepSWE v1.1, a test of long-horizon software engineering, compared with 74.2% for Anthropic's Claude Opus 5.5 and a tenth of a point further back for OpenAI's GPT-6 Astra. The agentic coding interface from Google Antigravity now runs directly inside the Gemini app for macOS, so a chat can become a coding session without switching tools. Google did not mention a Windows version.

The app release was part of a wider rollout. Koray Kavukcuoglu, Google's chief AI architect, wrote in a separate post that developers anywhere can now build on Argon through the paid tier of the Gemini API. Introductory rates of $2 per million input tokens and $10 per million output tokens last until Nov. 30, when they double to $4 and $20. Up to 10,000 scientists can apply for six months of free access to the highest usage tier in Antigravity, and cybersecurity teams that want Argon without its cyber guardrails still apply through the Fairwind Program.

On the consumer side, Argon is also available in AI Mode in Search for Pro and Ultra subscribers. A Google AI Pro subscription costs $19.99 a month in the U.S.; Ultra plans start at $99.99 and come with higher Argon limits. Google's plans page says tasks mode is limited to select countries for now and that Argon is still rolling out to Pro subscribers.

Engadget reported that Gemini's automation features extend beyond tasks mode. Users can ask Gemini to check their calendar every morning and estimate total travel and driving time using Maps, though Google may not account for real-time traffic because it uses data available when the answer is generated. They can schedule a daily news search, with a limit of 10 scheduled actions at a time, and receive a notification when results are ready. A weekly prompt can suggest different cafes and restaurants for the weekend if location access is granted; scheduled actions use the location accessed when the action was set up, not a current location while traveling.

Engadget also said users can define skills, which are replacing Gems, to remember preferences across conversations. A weekly newsletter writer, for example, can create a skill with brand guidelines and trigger it with a command such as /skillname; Google says the @ symbol will soon be used instead. For AI Pro and Ultra subscribers, Gemini Spark acts as an always-on agent that keeps working when phones and laptops are off. It can scan a week of emails, summarize the most important ones, create a to-do list and add appointments to a calendar. Gmail, Calendar and Drive must be connected first because they are off by default. Google said Spark is designed to check with users before taking major actions, and some Gemini features may be limited to specific regions.

Editor's Summary Google has moved Gemini 4 Argon from a limited cybersecurity release to Pro and Ultra subscribers in the Gemini app, adding a tasks mode that runs jobs while users are away and more than 40 connected apps. The rollout also extends Argon to AI Mode in Search and the Gemini API, while Engadget notes that automation tools such as scheduled actions, skills and Spark are expanding what the assistant can do outside a single chat.