Gemini CLI

What it does

Gemini CLI is an open-source terminal-based AI coding agent from Google that exposes Gemini 2.5 Pro/Flash and Gemini 3 models for interactive coding, codebase exploration, shell automation, and MCP tool use. It shares auth and quota with Gemini Code Assist, so the IDE plugin and CLI draw from the same daily budget.

Who it's for

Individual developers wanting a free, generous Gemini terminal experience; teams already on Google Workspace who can centralize on Code Assist Standard/Enterprise; and platform engineers integrating Gemini into pipelines via the Gemini API or Vertex AI.

How platform engineers use it

Scripted code review and refactor passes, log triage, runbook execution, infrastructure-as-code generation grounded in repo context, and pipeline integration via Gemini API key or Vertex AI service accounts. CI usage typically routes through the API or Vertex paths to bypass per-user daily caps and meet data-governance requirements.

Strengths

• Generous free tier (1,000 req/user/day with personal Google login)
• Open-source CLI; no vendor lock-in for the client itself
• Single auth/quota model shared with Gemini Code Assist IDE plugins
• Pay-as-you-go fallback via Vertex AI or Gemini API key when caps are hit
• Vertex AI path provides enterprise security, data governance, and zero-retention controls

Limitations

• Hard daily request caps on subscription tiers (no automatic burst)
• Free consumer tier uses prompts to improve Google products; not appropriate for proprietary code
• Model routing is opinionated and not always transparent
• Compliance posture varies by auth path (consumer vs Workspace vs Vertex AI)
• Subscription tiers above Pro require Workspace or Google AI Ultra purchase

AI maturity

Production-ready for individual developer use and Workspace-purchased team rollouts. Backed by Google's model and infrastructure investment, but the CLI itself is younger than competing agentic coding tools and the agent loop is still iterating rapidly.