If you want Claude Code but do not want to burn API credits, Google made that easier. I made a video installing Gemini CLI from scratch and putting it through a real task, and here is the gist plus the steps so you can follow along without watching if you prefer. It is an open-source coding agent that runs in your terminal, and its free tier is generous enough to actually work in. No credit card, no API key, just sign in with your Google account.
What the free tier gives you
The free tier is the headline. Sign in with a Google account and you get 1,000 requests per day, up to 60 requests per minute, and a 1 million token context window. For a lot of daily coding, that is enough to never touch a paid plan. It is licensed Apache 2, and out of the box it reads and edits your codebase, runs shell commands, fetches web pages, and grounds answers with Google Search built in. It also supports MCP, the same protocol Claude Code uses, so you can wire in GitHub, Slack, or a database.
There is a second way to run it free, using a Gemini API key, which suits developers who need specific model control. In the video I stick with the simpler Google-account path.
Setup takes about a minute
Install is a one-liner copied from the repo and pasted into your terminal. It asks how you want to authenticate, I pick log in with Google, approve, and it confirms the login. The CLI needs a restart, so press R, and you are in. In the walkthrough I do the whole thing from scratch so you can follow the exact steps.
If you want zero install, run it straight from npx.
Does it actually code
A free tier means nothing if the agent cannot build. So I gave it one prompt: create a simple snake game in this workspace. Gemini planned it, asked for permission at each step, wrote the code, and then told me how to run it. I opened the address it gave me and had a working snake game in a minute or two, all on the free tier. You can also set permissions to allow for the whole session so it stops asking.
By default it ran on the Gemini 3 model. Afterward I switched to manual model selection and picked the 3.1 Pro preview, one of the newer options, so you are not locked to a single model.
Try it
Run it now with no install: npx @google/gemini-cli
GitHub: https://github.com/google-gemini/gemini-cli
Watch the full walkthrough above, and subscribe for more on free and low-cost AI coding tools. I specialize in helping teams pick the right mix of AI coding agents for their budget and stack. Book a call at cloudyeti.io/meet.