Google has moved its Gemini-powered coding agent, Jules, into general availability after a two-month public beta, widening access to a tool the company bills as an "extra set of hands" for software development. Jules is built on the Gemini 2.5 Pro large-language model and operates asynchronously in a Google Cloud virtual machine. The agent can clone repositories, fix or update code, and open pull requests on GitHub while developers step away from their screens. New features rolling out with the launch include automatic pull-request creation and “Environment Snapshots” that store dependencies for faster, more consistent task execution. The service now follows a tiered pricing structure. An introductory free plan allows up to 15 individual tasks and three concurrent jobs each day. Heavier users can upgrade to Google AI Pro at $19.99 a month for roughly five times the limits, or to the Ultra tier at $124.99 a month for roughly 20 times the capacity. During the beta, thousands of developers submitted tens of thousands of tasks, leading to more than 140,000 code improvements shared publicly, according to Google. “The trajectory of where we’re going gives us a lot of confidence that Jules is around for the long haul,” Google Labs director of product Kathy Korevec said. The broader release of Jules is part of Google’s continuing effort to embed Gemini-based agents across its ecosystem, following recent launches of the Deep Think reasoning model and Guided Learning features aimed at students.
Comme ChatGPT, Gemini propose un mode spécial pour les étudiants https://t.co/GbeHih1G4j
Google has launched Gemini CLI GitHub Actions, an open-source AI agent designed to automate routine development tasks across repositories. Now in beta and globally available, this AI-powered assistant integrates seamlessly with @github , activating on key events such as new https://t.co/qLlLfQfc8O
Gemini’s Guided Learning mode can throw questions at you and whip up study tools you can use https://t.co/dn9z39p5RC