GC-101 · Module 2

Validation, Models & Sessions

3 min read

Gemini CLI gives you access to multiple Gemini models. You can switch between them via /settings. The two primary options are Gemini 2.5 Pro (the established workhorse, strong at code generation and reasoning) and newer models as Google releases them. Model selection should match your task — heavier reasoning tasks benefit from the most capable model, while routine operations work fine on faster variants.

# View and switch models
/settings
# Navigate to model selection

# Check current model and token usage
/stats model

Reasoning and thinking budget configuration controls how much "thinking" the model does before responding. Higher thinking budgets produce more thorough analysis but consume more tokens and take longer. For quick edits, keep thinking minimal. For architecture decisions or complex debugging, increase the budget. This is configured through /settings and gives you fine-grained control over the cost-quality tradeoff.

Session management in Gemini CLI is handled through /chat. You can view your session history, resume previous conversations, and share sessions as markdown or JSON exports. The ability to export sessions is particularly useful for documentation — when you solve a complex problem, export the session as a reference for your team.

# View session history
/chat

# Resume a previous session
/chat resume

# Share a session as markdown
/chat share --format markdown

# Export as JSON for programmatic use
/chat share --format json

Do This

  • Export sessions that solve complex problems as team documentation
  • Adjust thinking budget based on task complexity
  • Use /chat resume to continue work across terminal restarts

Avoid This

  • Let valuable debugging sessions disappear when you close the terminal
  • Use maximum thinking budget for every task regardless of complexity
  • Start fresh every time when /chat resume preserves your context