How to Use Gemini CLI: Complete Beginner Guide (2026)

Gemini CLI is Google's command-line tool for using Gemini models directly in your terminal. Like Claude Code, it lets you interact with AI without leaving the command line -- but it uses Google's Gemini models instead of Anthropic's Claude.

This guide covers installation, setup, your first task, and how Gemini CLI compares to other terminal AI tools.

In this guide:

• What Gemini CLI does
• Installation and setup (3 minutes)
• Your first Gemini CLI task
• Common use cases
• Gemini CLI pricing
• Gemini CLI vs Claude Code

Related guides: Gemini CLI vs Claude Code Comparison · How to Use Claude Code · How to Use Cursor AI · All AI Coding Tool Comparisons

What Gemini CLI Does

Gemini CLI brings Google's Gemini models to your terminal. You type natural language instructions, and Gemini CLI:

• Answers questions about code, APIs, documentation
• Writes and edits code in your project files
• Explains codebases by reading your project structure
• Runs shell commands when needed
• Processes files -- feed it documents, logs, or data for analysis

The key advantage: Gemini 2.5 Pro has a 1M token context window -- one of the largest available. This means Gemini CLI can process much larger files and codebases in a single conversation than tools using smaller-context models.

For a detailed comparison with Anthropic's CLI tool, see our Gemini CLI vs Claude Code comparison.

Installation and Setup

Prerequisites

• Node.js 18+ (check with node --version)
• A Google AI Studio API key (get one here)
• A terminal (Terminal on macOS, PowerShell on Windows)

Step 1: Install Gemini CLI

npm install -g @anthropic-ai/claude-code

Wait -- that's Claude Code. For Gemini CLI, the installation depends on the package:

npm install -g @anthropic-ai/gemini-cli

Actually, let's use the correct Google package. Install via:

npm install -g @google/generative-ai-cli

If npm isn't your thing, you can also use:

# Using the standalone installer
curl -fsSL https://cli.gemini.google.com/install.sh | bash

# Or with Homebrew (macOS)
brew install gemini-cli

Verify the installation:

gemini --version

Step 2: Set Your API Key

Get your API key from Google AI Studio. Google offers a free tier with generous limits (see pricing below).

Set the key as an environment variable:

macOS/Linux (add to ~/.bashrc or ~/.zshrc):

export GEMINI_API_KEY="your-api-key-here"

Windows PowerShell:

$env:GEMINI_API_KEY = "your-api-key-here"

Step 3: Start Using Gemini CLI

cd your-project-folder
gemini

You'll see the prompt:

gemini>

Total setup time: about 3 minutes.

Your First Gemini CLI Task

Example 1: Analyze Your Project

gemini> What does this project do? Summarize the tech stack and architecture.

Gemini CLI reads your project files and provides a summary. With the 1M token context window, it can process large codebases that other tools might struggle with.

Example 2: Write a Feature

gemini> Add pagination to the /api/users endpoint. Return 20 users per page with a total count. Include prev/next page links.

Gemini CLI finds the relevant file, reads the existing code, and adds pagination.

Example 3: Debug an Error

gemini> I'm getting a CORS error when calling the API from localhost:3000. Here's the error:

Access to fetch at 'https://api.example.com/users' from origin 'http://localhost:3000' has been blocked by CORS policy

Fix it.

Gemini CLI explains the CORS issue and provides the fix (adding appropriate headers to your server or configuring a proxy).

Example 4: Process a Large File

gemini> Read the file logs/app.log and find the most common error types. Summarize the top 5 errors with their frequency and likely causes.

This is where Gemini CLI's large context window shines -- it can process multi-megabyte log files that other tools might truncate.

Common Use Cases

Code Generation

gemini> Create a Express.js middleware that validates JWT tokens. Include:
- Token extraction from Authorization header
- Token verification using jsonwebtoken
- Error handling for expired and invalid tokens
- TypeScript types

Documentation

gemini> Read all files in src/api/ and generate API documentation in Markdown format. Include endpoint paths, HTTP methods, request/response schemas, and example curl commands.

Data Analysis

gemini> Read data/sales.csv and answer:
1. What's the total revenue by month?
2. Which product has the highest growth rate?
3. Are there any seasonal patterns?
Present the results in tables.

Refactoring

gemini> The file src/utils/helpers.ts has 15 exported functions that are used across the project. Split it into 3 files based on logical grouping (string utils, date utils, format utils). Update all imports across the project.

Gemini CLI Pricing

Google offers one of the most generous free tiers among AI providers:

Plan	Price	Rate Limits
Free	$0	15 requests/minute, 1M tokens/min, 1,500 requests/day
Pay-as-you-go	Per token	Higher rate limits, no daily cap

Pay-as-you-go pricing (Gemini 2.5 Pro):

• Input: $1.25 per 1M tokens
• Output: $10.00 per 1M tokens

Pay-as-you-go pricing (Gemini 2.0 Flash):

• Input: $0.10 per 1M tokens
• Output: $0.40 per 1M tokens

How this compares:

Model	Input (per 1M)	Output (per 1M)	Context Window
Gemini 2.5 Pro	$1.25	$10.00	1M tokens
Gemini 2.0 Flash	$0.10	$0.40	1M tokens
Claude 3.5 Sonnet	$3.00	$15.00	200K tokens
GPT-4o	$2.50	$10.00	128K tokens

Gemini is the cheapest per-token option with the largest context window. The trade-off: output quality on complex reasoning tasks is slightly below Claude Sonnet.

For a full pricing comparison, see our AI cost calculator and our 200-task benchmark report.

Gemini CLI vs Claude Code

Both are terminal-based AI coding tools, but they have different strengths:

Feature	Gemini CLI	Claude Code
Provider	Google	Anthropic
Model	Gemini 2.5 Pro	Claude 3.5 Sonnet
Context Window	1M tokens	200K tokens
Free Tier	Yes (generous)	$5 free credits
Input Price	$1.25/M tokens	$3.00/M tokens
File Editing	Limited	Full (reads, writes, creates)
Command Execution	Limited	Full (runs any terminal command)
Code Quality	Good	Very good
Large File Processing	Excellent	Good
Best For	Analysis, large codebases, free usage	Implementation, debugging, multi-file edits

When to Choose Gemini CLI

• You want a free terminal AI tool with generous limits
• You work with large files (log analysis, big data, long documents)
• You need the 1M token context window for processing entire codebases
• You want the cheapest per-token pricing

When to Choose Claude Code

• You need AI to edit files and run commands autonomously
• You want the highest code quality for complex implementations
• You need multi-step debugging with terminal command execution
• You want to connect to multi-agent squads via Ivern

For the detailed breakdown, see our Gemini CLI vs Claude Code comparison.

Using Both Together

Many developers use both: Gemini CLI for analysis and large-context tasks, Claude Code for implementation and debugging. Through Ivern Squads, you can coordinate both in the same workflow:

1. Gemini agent processes the codebase and identifies issues
2. Claude Code agent implements the fixes
3. A Reviewer agent verifies the changes

See our multi-agent coding setup guide for instructions on coordinating multiple terminal agents.

Frequently Asked Questions

How do I install Gemini CLI?

Install with npm (npm install -g @google/generative-ai-cli), set your API key from Google AI Studio as an environment variable (export GEMINI_API_KEY="your-key"), then run gemini in any project directory. Setup takes about 3 minutes. You need Node.js 18+ installed.

Is Gemini CLI free?

Yes. Google's free tier includes 15 requests per minute and 1,500 requests per day with Gemini 2.5 Pro. This is enough for most individual developers. Paid usage starts at $1.25 per million input tokens -- cheaper than Claude ($3.00/M) and GPT-4o ($2.50/M). See our cost calculator for monthly estimates.

What is the difference between Gemini CLI and Claude Code?

Gemini CLI uses Google's Gemini models with a 1M token context window and a generous free tier. Claude Code uses Anthropic's Claude models with stronger code generation and autonomous file editing. Gemini CLI is better for analysis and large-file processing. Claude Code is better for implementation and debugging. See our detailed comparison.

Can Gemini CLI edit files?

Gemini CLI can read and analyze files, but its autonomous file editing capabilities are more limited than Claude Code. For complex multi-file edits, Claude Code or Cursor are better options. Gemini CLI excels at analysis, explanation, and generation of code snippets that you can apply manually.

What context window does Gemini CLI support?

Gemini 2.5 Pro supports a 1M token context window -- roughly 750,000 words of text. This means Gemini CLI can process an entire medium-sized codebase in a single conversation, or analyze very large files (multi-MB logs, datasets, documentation) without truncation.

Can I use Gemini CLI for non-coding tasks?

Yes. Gemini CLI works for any text-based task: analyzing documents, summarizing reports, processing CSV data, generating documentation, writing emails, and more. The terminal interface and large context window make it particularly useful for processing large text files that won't fit in browser-based chat tools.

Next Steps

1. Install Gemini CLI: npm install -g @google/generative-ai-cli
2. Get your API key: Google AI Studio (free)
3. Try it: cd your-project && gemini
4. Compare tools: Gemini CLI vs Claude Code
5. Optional: Connect to Ivern Squads for multi-agent coordination

Related reading: Gemini CLI vs Claude Code · How to Use Claude Code · How to Use Cursor AI · AI Cost Calculator · All AI Coding Tool Comparisons

Start building with AI agent squads -- free signup