Google Launches AI Coding Agent Jules With Free and Paid Tiers

Google on 6 August officially moved its AI coding agent Jules out of beta, just over two months after a public preview began in May. The asynchronous tool, powered by the Gemini 2.5 Pro model, can clone codebases into Google Cloud virtual machines, open pull requests on GitHub and run self-directed bug-fixing or upgrade tasks while users work on other jobs. The general-availability release introduces a three-tier pricing structure. An introductory free plan allows up to 15 tasks a day and three to run concurrently. Paid access is folded into the company’s Google AI subscriptions: the Pro tier costs $19.99 a month for roughly five times the free limits, while the Ultra tier is priced at $124.99 a month and lifts capacity by about twenty-fold. Google says stability improvements and “hundreds” of interface tweaks made during testing underpin the wider rollout. During the beta, thousands of developers completed tens of thousands of jobs, contributing more than 140,000 publicly shared code improvements. The company also clarified that code from private repositories is excluded from AI-training datasets, responding to user feedback on data use. By running tasks in the cloud without requiring users to monitor sessions, Jules differentiates itself from synchronous coding assistants. Analysts expect the agent’s broader availability to intensify competition among AI developer tools—and, longer term, to reshape certain programming roles as automation gains ground.