Deprecated. Google announced at Google I/O 2026 (May 19) that Gemini CLI is being replaced by Google Antigravity — a closed-source, enterprise-oriented agent platform. Gemini CLI stops serving requests on June 18, 2026 for non-enterprise users (Gemini Code Assist Standard/Enterprise licenses retain access). Do not start new projects on Gemini CLI. Migrate to Antigravity CLI or another terminal agent.
Gemini CLI was Google's open-source terminal AI agent — powered by Gemini models and featuring a generous free tier. It accumulated 100,000+ GitHub stars and 6,000+ merged pull requests before being sunset in favour of Antigravity CLI.
Why It's in Hold
Gemini CLI has a confirmed deprecation date: June 18, 2026. All free-tier and Google AI Pro/Ultra users will stop receiving responses after that date. Enterprise Gemini Code Assist users retain access but new feature development has moved to Antigravity CLI. The open-source Apache 2.0 repository remains available but Google has signalled no intent to continue community contributions under the current model. The key policy change: Gemini CLI was open-source; Antigravity CLI is closed-source.
What to migrate to: Google Antigravity — Antigravity 2.0 includes an Antigravity CLI component that maintains Gemini CLI's core features (Agent Skills, Hooks, Subagents, Extensions-as-plugins) with Gemini 3.5 Flash as the default model.
Historical Context (retained for reference)
Original Capabilities
Gemini CLI runs a reason-and-act (ReAct) loop with built-in tools:
- File read/write operations
- Shell command execution
- Google Search grounding
- Web fetching
- MCP (Model Context Protocol) server support for custom integrations
# Install
npm install -g @google/gemini-cli
# Or run without installing
npx @google/gemini-cli
Plan Mode
A notable feature: Plan Mode puts the agent in read-only mode first, so it analyses your request, asks clarifying questions, and proposes a strategy before making any changes. This reduces the "AI ran off and broke everything" problem common with autonomous agents.
Advanced Access
- Google AI Studio or Vertex AI key: Usage-based billing, model choice flexibility
- Google AI Ultra subscription: Access to Gemini 3 Pro in the CLI
- Gemini Code Assist licence: VS Code integration + CLI unified under one plan
Google I/O 2026 — Chrome DevTools for Agents
Announced May 19, 2026: Chrome DevTools for agents gives Gemini CLI (and other coding agents including Google Antigravity, Cursor) direct access to a web app's runtime state — DOM, network activity, console output, and performance data — without the agent needing to parse screenshots or inject custom logging. This closes a persistent gap in agentic web debugging: agents previously had to guess at runtime behavior from static file content. The integration is surfaced via the DevTools Protocol; agents receive structured runtime data that can be referenced directly in multi-step debugging sessions.
Note: Free tier upgraded to Gemini 3 Flash as of the Google 3.x model family rollout (multiple secondary sources; verify against gemini-cli GitHub README for the exact model in use). Gemini 3.5 Flash (the new frontier Flash model) requires a paid Google AI plan.
Key Characteristics
| Property | Value |
|---|---|
| License | Open source (Apache 2.0) |
| Free tier | 1,000 req/day with Gemini 3 Flash |
| Install | npm install -g @google/gemini-cli |
| Provider | |
| GitHub | google-gemini/gemini-cli |
Gemini CLI is Google's open-source terminal AI agent — powered by Gemini models and featuring a generous free tier that makes it one of the most accessible agentic coding tools available.
Why It's in Trial
Launched mid-2025, Gemini CLI competes directly with Claude Code and OpenAI Codex CLI as a terminal-based agentic tool. Its standout feature is the free tier: log in with any Google account and get:
- 60 model requests per minute
- 1,000 requests per day
- Access to Gemini 2.5 Pro (1M token context)
- All at no cost
This makes it by far the most accessible entry point into agentic coding for engineers new to AI tools.
How It Works
Gemini CLI runs a reason-and-act (ReAct) loop with built-in tools:
- File read/write operations
- Shell command execution
- Google Search grounding
- Web fetching
- MCP (Model Context Protocol) server support for custom integrations
# Install
npm install -g @google/gemini-cli
# Or run without installing
npx @google/gemini-cli
Plan Mode
A notable feature: Plan Mode puts the agent in read-only mode first, so it analyses your request, asks clarifying questions, and proposes a strategy before making any changes. This reduces the "AI ran off and broke everything" problem common with autonomous agents.
Advanced Access
- Google AI Studio or Vertex AI key: Usage-based billing, model choice flexibility
- Google AI Ultra subscription: Access to Gemini 3 Pro in the CLI
- Gemini Code Assist licence: VS Code integration + CLI unified under one plan
Key Characteristics
| Property | Value |
|---|---|
| License | Open source (Apache 2.0) |
| Free tier | 1,000 req/day with Gemini 2.5 Pro |
| Install | npm install -g @google/gemini-cli |
| Provider | |
| Announcement | Introducing Gemini CLI |
| GitHub | google-gemini/gemini-cli |