Gemini API developers
Use a terminal agent with explicit model selection and a direct Gemini API-key billing path.
Independent tool overview
Gemini CLI at a glance
Gemini CLI is Google's Apache-licensed terminal agent for understanding codebases, editing files, running commands, searching the web, and connecting external tools through MCP.
Visit the official Gemini CLI site ↗
- Product type
- Open-source terminal coding agent
- Developer
- Google and open-source contributors
- License
- Apache 2.0
- Package
- @google/gemini-cli
- Personal Google login
- Discontinued
- Supported billing routes
- Gemini API, Vertex AI, or organization license
- Free API-key quota
- 250 requests per day, Flash only
- Official personal successor
- Antigravity CLI
- Last reviewed
- August 29, 2026
Overview
What Gemini CLI is
Gemini CLI is an open-source terminal agent that can inspect repositories, edit multiple files, run shell commands, search or fetch the web, and work interactively or in scripts. The command-line client is licensed under Apache 2.0.
The project supports Gemini models with long context, persistent GEMINI.md project instructions, resumable sessions, plan mode, extensions, MCP servers, custom commands, skills, hooks, GitHub workflows, and IDE integration.
The most important current caveat is authentication. On June 18, 2026, Google stopped serving Gemini CLI requests for Gemini Code Assist for individuals, Google AI Pro, and Google AI Ultra, and removed personal 'Login with Google' access. Google directs those users to Antigravity CLI.
Gemini CLI remains relevant for teams using a Gemini API key, Vertex AI, or Gemini Code Assist Standard or Enterprise. The open-source repository also remains active, but the right setup now depends on whether the user wants the supported Google successor or specifically needs the existing Node-based client.
Like every coding agent with shell and file tools, Gemini CLI can make incorrect edits or execute risky commands. Use a clean branch, review its plan and diff, keep approvals enabled, run it in a sandbox where practical, and never give an MCP server or extension more access than the task requires.
Use cases
Who Gemini CLI is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Google Cloud teams
Run through Vertex AI or an assigned Gemini Code Assist Standard or Enterprise license.
Open-source experimentation
Inspect, extend, contribute to, or self-package an Apache-licensed coding-agent client.
Large-codebase analysis
Use Gemini's long context with repository search, file operations, and persistent project instructions.
Terminal automation
Run prompts non-interactively and consume text, JSON, or streaming JSON output in scripts and workflows.
Extensible agent workflows
Add MCP servers, extensions, commands, skills, hooks, and IDE or GitHub integrations.
Capabilities
Core Gemini CLI features
Repository understanding
Reads and searches project files to explain architecture, trace behavior, and answer codebase questions.
Multi-file editing
Creates and modifies code across a workspace, subject to the selected approval and policy settings.
Shell tools
Runs tests, builds, linters, package commands, and other terminal operations.
Web search and fetch
Grounds research with Google Search and reads web resources when tools and policy allow.
Project context
Loads GEMINI.md files for persistent repository instructions, conventions, commands, and background.
Plan mode
Restricts the agent to read-only research and planning tools until the user approves implementation.
Sessions and rewind
Saves, resumes, and revisits conversations for longer coding work.
MCP support
Connects external tools and data sources through configured Model Context Protocol servers.
Extensions and custom commands
Packages reusable MCP integrations, configuration, context, slash commands, and other project capabilities.
Headless mode
Runs non-interactively with text, JSON, or streaming JSON output for automation.
Sandbox and policy controls
Supports sandboxed execution, trusted-folder behavior, approval modes, and enterprise-admin restrictions.
GitHub and IDE integration
Supports automated issue or pull-request workflows and a companion IDE experience.
Process
How the Gemini CLI workflow works
Step 1
Choose Gemini CLI or its successor
Personal Google-account users should start with Antigravity CLI; use Gemini CLI only when its supported API, Vertex, or organization route matches the requirement.
Step 2
Install the stable package
Run it through npx or install the latest stable npm, Homebrew, MacPorts, or supported environment package.
Step 3
Configure supported authentication
Use a Gemini API key, Vertex AI credentials, or a Gemini Code Assist Standard or Enterprise license; do not depend on the discontinued personal Google login.
Step 4
Open a low-risk repository
Begin on a clean branch with recoverable changes, working tests, and no unrelated secrets or sensitive files in scope.
Step 5
Create project instructions
Add a focused GEMINI.md with architecture, conventions, required commands, boundaries, and definition of done.
Step 6
Enable safety controls
Use plan mode for discovery, retain confirmations, configure policies, and enable sandboxing for command execution where practical.
Step 7
Request a bounded task
Specify the exact outcome, files or modules in scope, validation commands, and actions that require confirmation.
Step 8
Review every action
Inspect proposed commands and tool requests, especially network calls, package changes, credentials, and write-enabled MCP actions.
Step 9
Validate the result
Review the diff, run tests and builds, examine security-sensitive changes, and keep normal pull-request protections.
Step 10
Monitor quota or spend
Use the model statistics command and provider billing controls to track requests, tokens, and pay-as-you-go usage.
Cost
Gemini CLI pricing and free plan
The Gemini CLI client is free and open source, but model access follows the selected authentication provider. Personal Login with Google no longer works for Gemini CLI. Free and paid API routes have different model access and quotas, while organization licenses use per-user daily request limits.
Gemini CLI software
Free
The command-line client is available under Apache 2.0.
- No software license fee
- Model access is billed or limited separately
- Personal Google-account authentication is discontinued
- Users remain responsible for infrastructure and integration costs
Gemini API key free tier
$0 within quota
For light experimentation through the unpaid Gemini API tier.
- 250 maximum model requests per user per day
- Flash model only
- Per-minute limits also apply
- Availability is not guaranteed at peak demand
Vertex AI Express Mode
Free trial path
A temporary Vertex AI entry route without initially enabling billing.
- Available for 90 days before billing is required
- Quotas and models vary by account
- Google Cloud setup is required
- Move to regular Vertex AI for ongoing usage
Gemini API or Vertex AI
Pay as you go
For model selection, higher usage, production workloads, or uninterrupted access.
- Pricing varies by model and token usage
- Quotas vary by tier or Google Cloud allocation
- Vertex AI supports enterprise governance and Cloud integration
- Use provider budgets and usage monitoring
Gemini Code Assist Standard
Paid organization seat
An organization license that continues to support Gemini CLI.
- 1,500 maximum model requests per user per day
- Purchased and assigned through Google Cloud
- Model routing is managed by Gemini CLI
- Fixed-seat pricing depends on subscription terms
Gemini Code Assist Enterprise
Paid enterprise seat
Higher-quota organization access with enterprise features.
- 2,000 maximum model requests per user per day
- Gemini CLI access remains supported
- Enterprise administration and governance
- Pricing depends on subscription terms
Pricing checked . Check current pricing at the source ↗
Assessment
Gemini CLI strengths and limitations
Where it stands out
- Apache 2.0 client can be inspected, extended, and contributed to
- Strong terminal-first codebase understanding and editing workflow
- Long-context Gemini models help with large repositories and multimodal inputs
- Built-in file, shell, web-search, and web-fetch tools cover common development tasks
- Plan mode provides a useful read-only step before implementation
- MCP, extensions, skills, hooks, commands, and GEMINI.md make the agent highly customizable
- Headless JSON and streaming output support automation
- Multiple provider routes cover API experimentation and Google Cloud deployments
- Enterprise administrators can restrict authentication, extensions, MCP, approvals, and high-risk modes
What to consider
- Personal Login with Google stopped working on June 18, 2026
- Google directs individual and Google AI Pro or Ultra users to Antigravity CLI
- The repository README and older articles may still advertise personal free-login limits that no longer reflect the supported consumer path
- The free Gemini API-key tier is limited to 250 requests per day and Flash models
- Pay-as-you-go pricing varies by model and token usage, making costs task-dependent
- Model requests, file reads, and tool outputs can send sensitive context to the configured model provider
- Shell commands and file writes can damage a working tree or local environment if approved without review
- Sandboxing and policy behavior must be configured and tested for the deployment environment
- YOLO-style approval removes important safety checks and is unsuitable for sensitive repositories
- MCP servers and extensions add their own permissions, supply-chain risk, and data handling
- Usage statistics are enabled unless disabled in privacy settings
- Model-generated code can contain bugs, insecure patterns, or fabricated assumptions
- Fast release channels can introduce regressions; production teams should use the stable package
- A terminal agent does not replace tests, code review, secret scanning, or branch protections
Compare
Gemini CLI alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Coding
Antigravity
Choose Antigravity for Google's supported successor experience, especially if you previously used Gemini CLI with a personal Google account.
Explore Antigravity →Coding
Claude Code
Consider Claude Code for another mature terminal-first coding agent with a strong repository workflow.
Explore Claude Code →Coding
Codex App
Choose Codex when you prefer OpenAI's coding agents and a dedicated interface for coordinating multiple tasks.
Explore Codex App →Questions
Gemini CLI FAQs
What is Gemini CLI?
Gemini CLI is Google's open-source terminal agent for reading and editing code, running commands, searching the web, and connecting tools through MCP.
Is Gemini CLI still active?
The open-source project remains active, and Gemini CLI continues to support Gemini API, Vertex AI, and Gemini Code Assist Standard or Enterprise routes. Its personal Google-account access was discontinued.
Can I still sign in to Gemini CLI with a personal Google account?
No. Google stopped serving Gemini CLI requests for Gemini Code Assist for individuals, Google AI Pro, and Google AI Ultra on June 18, 2026 and removed the personal Login with Google path.
What replaced Gemini CLI for individual users?
Google directs affected consumer users to Antigravity CLI, the terminal product in its newer Antigravity agent family.
Is Gemini CLI free?
The client itself is free under Apache 2.0. Model usage may be free within an eligible API quota, included with an organization license, or billed by the Gemini API or Vertex AI.
What is the current Gemini CLI free quota?
The official quota page lists 250 requests per user per day for an unpaid Gemini API key, limited to Flash models. Vertex AI Express Mode has account-specific limits and a 90-day no-billing period.
Does Gemini CLI support MCP?
Yes. It can discover and call tools from configured MCP servers, and extensions can bundle MCP configuration. Review every server's code, credentials, permissions, and data handling.
Can Gemini CLI run in scripts?
Yes. Headless mode supports plain text, JSON, and newline-delimited streaming JSON output for automated workflows.
Does Gemini CLI have a sandbox?
Yes. It supports sandboxed tool execution and permission expansion, but teams should verify the active configuration rather than assume every command is isolated.
Does Gemini CLI collect telemetry?
The project documents anonymized usage statistics covering tool names, request timing, success status, and configuration—not prompt, response, or file contents. Users can disable usage statistics in privacy settings.
Bottom line
Our Gemini CLI verdict
Gemini CLI remains a capable, highly extensible open-source agent for API and Google Cloud users, but it is no longer the default choice for individuals who want to sign in with a Google account. Those users should start with Antigravity CLI. Teams that deliberately keep Gemini CLI should use a supported authentication route, stable releases, plan mode, sandboxing, tightly scoped tool permissions, and normal code-review controls.
Visit Gemini CLI website ↗Similar Coding Tools
View all Coding AI tools →
Warp 2.0
Warp 2.0: All-in-one agentic development platform
Bolt.new
Bolt.new: Lets you spin up AI-powered web apps instantly using natural language.
Google Firebase
Google Firebase: Firebase helps developers build and scale apps with backend infrastructure, analytics, and real-time databases.
Cursor Agents
Cursor Agents - Work with a powerful coding assistant anywhere with the Cursor Agent on web and mobile
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.