Google Launches Gemini Tasks and Gemini 4 Neon Model
Google released Gemini Tasks for free in six countries and introduced the Gemini 4 Neon AI model for coding and analysis.
Google released Gemini Tasks for free on November 17, 2026, to users in the United States, India, Japan, South Korea, Canada, and Australia. Powered by the Gemini 4 Hydrogen model, the feature—formerly known as Gemini Spark—allows the AI to act as an agent to handle multi-step projects, such as booking service appointments via Google Duplex, analyzing utility bills, and shopping for groceries.
In the United States, the service includes integrations for RCS texting and email. Financial transactions are processed through Google Pay using the Agent Payments Protocol in collaboration with Mastercard and Visa, with American Express support expected soon. To maintain security, Google runs tasks in hardware-isolated virtual machines and requires human approval for sensitive actions.
On November 18, the company expanded its AI offerings by launching the Gemini 4 Neon model. Positioned as a mid-tier option between the Hydrogen and Carbon models, Neon features a 1 million-token context window and four adjustable thinking levels for coding and knowledge work. It is available to developers via the Gemini API and to AI Pro, Ultra, and enterprise subscribers.
Other updates include the global launch of Temporary Threads in AI Mode for Search, which prevents queries from being saved to account history, and the expansion of personal avatars in Google Vids to all personal account holders. Josh Woodward, VP of Product for AI, also announced updates to improve tool call reliability within the Gemini and Google Workspace integration.