On this page
Gemini CLI Commands Reference: What Still Works After Jun 18
As of June 18, 2026, free and paid consumer Gemini CLI access ended, with only enterprise licenses retaining functionality. This reference covers the full v0.47.0 command set, including slash commands, CLI flags, and migration details for users evaluating alternatives.
Google’s open-source terminal agent proved that developers want AI in their command line, not just another browser tab. Then the company pulled the rug out. As of June 18, 2026, Gemini CLI stopped serving requests for free-tier, Google AI Pro, and Google AI Ultra users, forcing a migration to the closed-source Antigravity CLI. If you’re on an enterprise license, your access continues. Everyone else needs to understand what they’re losing — and what commands still matter if you’re evaluating alternatives.
This reference covers the full command set for Gemini CLI v0.47.0, the last stable release before the consumer shutdown. Whether you’re auditing your existing workflows or comparing capabilities against Claude Code’s 90+ slash commands, this is the analytical breakdown you need.
What Gemini CLI Actually Does
Gemini CLI is an open-source AI agent for the terminal, released under the Apache 2.0 license. It runs on macOS, Linux, and Windows, and requires Node.js 20 or newer to function. You can install it via npx, npm global install, Homebrew, MacPorts, or Anaconda — pick whichever fits your setup.
The tool gives you a 1 million token context window when using the Gemini 3 model family, which is five times larger than Claude Code’s 200K limit. That matters when you’re feeding it large codebases or complex refactoring tasks. Built-in tools cover reading and editing files, running shell commands, web fetching, Google Search grounding, and GitHub-style tasks like pull request review and rebase help, per OpenTools.
It also supports the Model Context Protocol (MCP) for connecting to external servers and custom tools. If you’ve already read our MCP developer guide, you know the tradeoffs there — higher costs, lower reliability than direct CLI integration for most teams.
Authentication Tiers and Quotas
Your access level determines how many requests you get per day. The authentication methods for Gemini CLI include personal Google account, Gemini API key, and Vertex AI, according to the official quota documentation.
Here’s the breakdown:
| Authentication Method | Tier | Max Requests/Day |
|---|---|---|
| Google Account | Gemini Code Assist (Individual) | 1,000 |
| Google Account | Google AI Pro | 1,500 |
| Google Account | Google AI Ultra | 2,000 |
| Gemini API Key | Free Tier (Unpaid) | 250 (Flash only) |
| Google Workspace | Code Assist Standard | 1,500 |
| Google Workspace | Code Assist Enterprise | 2,000 |
| Google Workspace | Workspace AI Ultra | 2,000 |
The free tier for personal Google accounts provides 1,000 maximum model requests per user per day and 60 requests per minute. Users authenticating with an unpaid Gemini API key get 250 requests per day, restricted to the Flash model. Google AI Pro subscribers using Gemini CLI have a quota of 1,500 requests per user per day, while Ultra subscribers get 2,000.
As of June 18, 2026, Gemini CLI stopped serving requests for free-tier, Google AI Pro, and Google AI Ultra users, per Google’s deprecation notice. Gemini Code Assist Standard and Enterprise subscribers retain full access. If you’re not on an enterprise license, these quotas are now theoretical — you’ll need to migrate to Antigravity CLI or switch tools entirely.
Core Slash Commands You Should Know
The slash command interface is where Gemini CLI diverges from simple chat wrappers. These aren’t just shortcuts — they’re workflow primitives.
The /resume command is the canonical command for browsing, saving, resuming, and managing past sessions, with /chat as an alias, per Toolsbase. You can list, save, resume, delete, share, and create checkpoints from previous conversations. This is critical for long-running projects where context spans multiple sessions.
/restore reverts files to their state before the last tool call. /rewind rewinds the conversation history to a previous turn. Together, these give you an undo stack that most coding agents lack — you can experiment aggressively and roll back without manual git gymnastics.
The /memory command provides an inbox for reviewing and patching extracted skills, per the v0.39.0 commit. This is how Gemini CLI persists learned patterns across sessions, and it’s one of the features that doesn’t have a clean equivalent in Antigravity CLI at launch.
Other essential slash commands include /tools, /mcp, /hooks, and /plan, all documented in the Toolsbase reference. These let you inspect available tools, manage MCP server connections, configure hooks for automation, and toggle plan mode for complex multi-step tasks.
CLI Flags for Automation and Scripting
If you’re wiring Gemini CI into CI/CD pipelines or shell scripts, the CLI flags matter more than the interactive commands. Here’s what you need:
--session-fileloads a session from JSON;--session-idstarts a session with a fixed UUID--list-sessionsand--delete-sessionmanage sessions programmatically--policyand--admin-policyenable Policy Engine integration for enterprise governance--acpenables stable ACP mode for agent communication protocols--raw-outputand--accept-raw-output-riskenable raw output mode for machine parsing--output-formatsupports JSON and stream-json output--worktreeor-wenables git worktree integration
The --experimental-zed-integration flag was removed in v0.47.0, so don’t waste time looking for it. Session management flags are particularly important if you’re running Gemini CLI in headless mode for automated code review or testing pipelines.
Keyboard Shortcuts and Editor Integration
The keyboard shortcut Ctrl+G opens an external editor, and F4 shows IDE context, per pull request #24861. Ctrl+X is deprecated for opening an external editor — users are prompted to use Ctrl+G instead. This change shipped in April 2026 and reflects Google’s push toward IDE-like interaction patterns even in the terminal.
These shortcuts matter less now that consumer access is gone, but if you’re on an enterprise license, they’re part of the daily workflow. The external editor integration lets you compose complex prompts in your preferred editor rather than fighting with terminal input limitations.
The Migration Reality Check
To migrate from Gemini CLI to Antigravity CLI, users should run agy plugin import gemini, rename GEMINI.md to AGENTS.md, and move .gemini/skills/ to .agents/skills/, per the Developers Digest migration guide. Antigravity CLI uses a weekly compute-based quota model instead of Gemini CLI’s daily request caps, which heavy users are finding more restrictive.
The MCP configuration in Gemini CLI is inline in settings.json, while Antigravity CLI uses a separate mcp_config.json. That’s a breaking change for anyone with complex MCP setups. If you’re evaluating whether to migrate or switch to a competitor, the Claude Code vs Gemini CLI comparison breaks down the updated pricing and capability differences.
Google’s decision to leverage thousands of unpaid community contributions to build Gemini CLI, then retire the open-source version for all non-paying users while retaining it for enterprise customers, is a deliberate user segmentation strategy. The open-source repository remains Apache 2.0 licensed, but without API access, that license is meaningless for the vast majority of the original contributor base. Contributors invested labor under the assumption their work would benefit the broader community, not exclusively paying enterprise clients.
What This Means for Your Workflow
If you’re an enterprise user, your Gemini CLI access continues unchanged. Audit your dependencies anyway — the ecosystem support will decline as the community moves to Antigravity or competing tools.
If you’re a consumer user, you have three realistic options: migrate to Antigravity CLI and accept the closed-source tradeoffs, switch to Claude Code or another competitor, or self-host against a paid Gemini API key. Each path has different cost structures and capability gaps.
The command reference above still matters because it lets you compare feature parity honestly. Don’t assume the replacement matches what you’re losing — verify it against your actual workflow.