
Gemini CLI
Free, open-source terminal agent from Google that brings Gemini 2.5 Pro with a 1M-token context window to your command line.

Short description
Gemini CLI is Google’s open-source, terminal-based AI coding agent that brings Gemini models directly into your development workflow. It can understand large codebases, edit files, run commands, and automate multi-step coding tasks without requiring a separate AI-powered IDE. Its generous free tier makes it an accessible way to experiment with Gemini-powered coding, while API keys and Vertex AI provide alternative access options
Features
1M-token context: Reads your entire codebase in one shot (Gemini 2.5 Pro) instead of windowing it
Multimodal generation: Generates apps from PDFs, images or sketches you paste in
Google Search grounding: Pulls real-time information from Google Search into its answers
MCP support: Connects to Model Context Protocol servers (including Imagen/Veo/Lyria media generation)
Conversation checkpointing: Saves and resumes long sessions where you left off
Gemini md: Project context file that tailors its behavior per repository
GitHub Action: PR reviews, issue triage and @gemini-cli mentions inside your repo
Pricing
Free: 60 requests/min and 1,000 requests/day with a personal Google account (Gemini 2.5 Pro, 1M-token context)
Optional Gemini API key: 100 requests/day free
Vertex AI: Google Cloud access for enterprise deployments, with usage-based pricing.
Open source: Apache-2.0 license; no fee to download or run the CLI itself.
Integrations
MCP servers: Connects external tools and media generation
Google Search grounding: Real-time web information
GitHub Action: Works inside your repo
Vertex AI & AI Studio: Google’s model platforms
Gemini models: 2.5 Pro and Flash
Platforms
- Terminal/CLI (macOS
- Linux
- Windows; Node 20+) and non-interactive mode for scripts
What users love about Gemini CLI
Free usage: Free with a personal Google account, with 60 requests/min and 1,000 requests/day on Gemini 2.5 Pro
Large context window: The 1M-token context lets it read an entire codebase in one go instead of in pieces
Build apps from a sketch: You can paste in PDFs, images or sketches and it generates an app from them
What users complain about Gemini CLI
No graphical interface: It is terminal-only, so developers who prefer an IDE have no editor to work in
Only Gemini models: You can’t switch to Claude or GPT
Free-tier limits under heavy use: The 60 requests/min and 1,000 requests/day caps can run out when you use it heavily
Similar tools
Alternative AI options