gemini

Start an interactive session

$ gemini

$ gemini -p "[explain this code]"

$ gemini -m [gemini-2.5-flash]

$ gemini -p "[prompt]" --output-format json

$ gemini --sandbox

$ gemini``` then ```@[path/to/file.py] [explain this code]

$ gemini``` then ```![ls -la]

gemini [options] [prompt]

Gemini CLI is Google's open-source AI agent that brings the power of Gemini directly into your terminal. It enables developers to write, refactor, and debug code, execute commands, and analyze entire repositories using natural language.
The CLI uses a reason and act (ReAct) loop with built-in tools including Google Search grounding, file operations, shell commands, and web fetching. It supports MCP (Model Context Protocol) for custom integrations and can be configured with project-specific instructions via GEMINI.md files.
Free tier users get 60 requests per minute and 1,000 requests per day with a personal Google account, accessing models with a 1 million token context window. The tool works across macOS, Linux, and Windows.

-p PROMPT

Run in non-interactive mode with the given prompt

Specify model variant (gemini-2.5-pro, gemini-2.5-flash, gemini-3-pro)

Output format: text, json, or stream-json

Run in sandboxed Docker environment for safety

Auto-approve all actions (enables sandbox by default)

Display help information

Show version number

/model

Select a different Gemini model

Clear terminal screen and scrollback

Browse and resume previous conversations

Show token usage and session duration

Summarize chat context to save tokens

Add text to AI's instructional memory

List available tools

Show connected MCP servers

Exit the CLI

Include file or directory content in prompt

Execute a shell command

Requires authentication with a Google account or API key (set via GEMINI_API_KEY environment variable). Shell commands and file modifications can be dangerous; use --sandbox mode for untrusted operations. Token usage can be significant for large codebases.

Gemini CLI was launched by Google on June 25, 2025 as an open-source tool under the Apache 2.0 license. It was built on the Gemini 2.5 Pro model and introduced alongside a new Gemini logo. By December 2025, free tier users gained access to Gemini 3 Pro and Flash models.

claude(1), opencode(1), aider(1), gcloud(1)