Google opens Gemini autonomous tasks feature for free across six countries
Google AI|AIDEN Editorial|Nov 17, 2026|Updated Nov 21, 2026
Google has expanded its Gemini Tasks agent feature to general users free of charge across six countries, including South Korea and the United States. Previously restricted to Gemini Advanced subscribers, the multi-step task execution tool is powered by a new lightweight model designed to slash inference costs. In the United States, the rollout also activates advanced integrations enabling the AI agent to make phone bookings and process payments.
Google began rolling out its autonomous Tasks capability within the Gemini mobile and web applications free of charge on Nov. 17, 2026, across six countries: South Korea, the United States, Japan, Canada, Australia, and India. The agentic tool, which was previously restricted to paid Gemini Advanced subscribers, is now available to all standard account holders aged 18 and older in these launch markets. Alphabet and Google chief executive officer Sundar Pichai announced via social media that the free rollout targets hundreds of millions of users across the six nations. This expansion signals a pivotal shift from conversational answer generation toward autonomous multi-step execution, allowing the artificial intelligence system to handle end-to-end assignments such as scheduling coordination and document organization.
The widespread rollout is enabled by a strategic backend model transition aimed at curbing operational expenses. While earlier subscription testing relied on expensive Gemini Pro tier models, the free tier is anchored by Gemini 4 Hydrogen, a lightweight architecture officially launched on Nov. 12, 2026. Bypassing the flagship Gemini 4 Carbon, Google leveraged Hydrogen at pricing levels of $1.50 per million input tokens and $7.50 per million output tokens, substantially lowering inference infrastructure costs during massive call volumes. This dramatic reduction in foundational model operating costs serves as the primary technical catalyst for delivering scalable, free autonomous agent services to broad consumer audiences.
As the agent transitions beyond digital processing into physical-world transactions, Google has introduced deeper real-world action capabilities alongside strict safety guardrails. Josh Woodward, vice president for Google Labs and the Gemini app, stated that Gemini can now make calls, process payments, and finalize assignments for users. In the United States, the deployment activates Google Duplex calling for store and repair shop reservations, Agent Payments Protocol support via Google Pay, and dedicated text and email handling. To mitigate risks associated with expanded user permissions, each active session runs inside an account-dedicated hardware-isolated virtual machine, accompanied by prompt-injection filters and mandatory user-configured spending thresholds. In South Korea, eligible users can now toggle from chat to Tasks mode to automate Google Calendar and Google Drive workflows without subscription fees, while specialized calling and payment features remain slated for future expansion as market competition shifts toward autonomous execution.
— AIDEN Editorial Team
What this means for the market
The aggressive distribution of free, multi-step autonomous agents marks a broader structural transition across the generative artificial intelligence industry from passive conversational bots to action-oriented execution engines. By anchoring large-scale deployment to cost-efficient lightweight models, major platform operators demonstrate that sustainable inference economics are the true prerequisite for mass agent commercialization. As these software agents gain the authority to execute real-world financial transactions and telephone interactions, enterprise developers and global regulators will increasingly focus on hardware-level execution boundaries, prompt-injection defense frameworks, and consumer liability rules.
How this issue is unfolding
Competition across the generative artificial intelligence sector is shifting rapidly from conversational chatbots to autonomous agents capable of completing multi-step tasks on behalf of users. Following its initial introduction under the Gemini Spark name at Google I/O in May and subsequent testing for paid subscribers, Google made the bold move to offer the Tasks capability entirely free of charge to hundreds of millions of regular users across six countries. This expansion strategy was made viable by adopting the highly inference-efficient Gemini 4 Hydrogen model as the backend infrastructure rather than a high-cost flagship, successfully driving down the expense of large-scale agent deployment. Furthermore, as artificial intelligence acquires physical and economic execution capabilities, such as booking automotive service appointments over the telephone and processing payments via credit cards, establishing secure isolated computing infrastructure and spending control safeguards has emerged as a central issue for commercialization.