Jules by Google
#41 in Developer Tools
« Automate code tasks such as bug fixing, test writing and dependency updates via an asynchronous AI agent integrated into GitHub »
Jules by Google: hand off a bug, get back a ready-to-review pull request
Updated on September 15, 2026
Jules by Google is an autonomous coding agent that clones your GitHub repository into a secure cloud virtual machine, fixes bugs, writes tests and comes back with a pull request ready for review. First shown in late 2024 inside Google Labs, it left beta in August 2025 and runs asynchronously, so your editor never waits. The free tier covers 15 tasks a day, with far higher ceilings on the Google AI Pro and Ultra subscriptions. There is nothing to install, everything happens in the browser.
Pros
- Works in the background while you code elsewhere
- Editable plan before and during execution
- Opens a GitHub pull request once you approve the diff
- 15 free tasks daily, no card required
- CLI, public API and GitHub label triggers
Cons
- Limited to GitHub repositories for now
- Free tier runs an older Gemini generation
- Large refactors can take several passes
Jules clones your repo and works while you eat lunch
Jules copies your code into an isolated virtual machine on Google Cloud, writes a plan, edits files across the codebase and opens a pull request on GitHub. You read the diff, merge, done.
The routine is simple. You paste a bug description before lunch, tweak the proposed plan if one step looks off, and the pull request is waiting when you get back, browser notification included. Since the beta phase, over 140,000 code changes produced by the agent have been shared publicly.
- Test writing and bug fixing
- Feature additions and small migrations
- Dependency version bumps
- Audio recaps of recent commits
Hands-on with Google's free async coding agent
Jules debugging a multiplayer game end to end
Terminal, API, GitHub labels, Jules takes orders from anywhere
Jules Tools, a lightweight command-line companion installed through npm, starts and monitors tasks without leaving the terminal. A public API plugs the agent into CI/CD pipelines, GitHub Actions, Slack, Linear or Jira, and it remembers your preferences repo by repo.
There is a neat shortcut too. Adding a jules label to a GitHub issue assigns the work directly (handy for clearing a backlog on a Friday evening). This cloud-based, fire-and-forget format places it among the autonomous coding agents alongside OpenAI Codex and Devin, far from the inline completion style of Cursor or GitHub Copilot.
Fifteen free tasks a day, then the Google AI plans
Jules has no standalone subscription, its limits follow your Google AI plan. The free tier asks for no payment details and covers a decent daily batch of chores.
Google AI Pro costs $19.99 per month, and Ultra starts at $99.99. Since these figures move around, the official Jules limits page is the one to check before upgrading.
| Plan | Tasks per day | Concurrent | Model |
|---|---|---|---|
| Jules free | 15 | 3 | Gemini 2.5 Pro |
| Jules in Google AI Pro | 100 | 15 | Gemini 3 Pro |
| Jules in Google AI Ultra | 300 | 60 | Gemini 3 Pro |
Frequently asked questions
Is Jules by Google free?
Yes, Jules has a permanent free tier with 15 tasks per day and 3 concurrent tasks, no payment card required. Google AI Pro and Ultra subscriptions raise the ceilings to 100 and 300 daily tasks respectively, and switch the agent to a newer Gemini model than the free version.
Jules or OpenAI Codex, which one should you pick?
Both agents clone your repository in the cloud and deliver a pull request. Codex comes bundled with a ChatGPT subscription, while Jules rides on the Google AI plans and keeps a genuine free tier of 15 daily tasks, a rarity in this category. If your stack already leans on Google services, Jules slots right in.
Does Google train its models on my code through Jules?
No, Google states that Jules does not train on your data. Every task runs inside a temporary, isolated Google Cloud virtual machine that is discarded once the work is delivered. Human review of the diff before merging remains essential, as with any coding agent.
Which programming languages does Jules handle?
Node.js, Python, Go, Java and Rust are among the supported languages, and the agent is largely language-agnostic since it reads the full project context. JavaScript and TypeScript are common picks as well. Environments can be configured with install scripts saved as reusable snapshots between tasks.
Verdict: Fifteen free tasks a day is a rare deal in this category, enough to delegate real work from day one. Developers who live on GitHub and would rather review pull requests than babysit a terminal will find a quiet, reliable teammate in Jules.
★ Featured AI Tools ★
AI Alternatives for
Jules by Google
Paid
Free
Free
Free
Free
Freemium
Free
