When should I enable 'Thinking' mode in Claude Code?

Enable Thinking mode for architectural decisions with complex trade-offs, deep debugging scenarios, and any task with a complexity score higher than 7/10.

What is the 20/60/20 rule for Claude models?

It is a strategic distribution recommending 20% Opus for complex tasks, 60% Sonnet for standard development, and 20% Haiku for simple operations to balance quality and cost.

When should I escalate from Sonnet to Opus?

Escalate to Opus when task complexity exceeds 8/10, when working on security-sensitive code, or for production-critical architectural decisions.

How much can I save by using Claude Haiku for simple tasks?

Using Haiku for formulaic tasks like documentation, simple file operations, and test generation can save approximately 80% in token costs compared to Sonnet.

Model Selection Guide

Name: Model Selection Guide
Author: Zate

byZate

Productivity & Workflow

Provides strategic guidance for choosing between Claude Opus, Sonnet, and Haiku models based on task complexity and token efficiency.

About

The Model Selection Guide is a specialized Claude Code skill designed to optimize development workflows by matching specific tasks to the most appropriate AI model. It provides a systematic framework for balancing speed, cost, and output quality, recommending a 20/60/20 distribution across Opus, Sonnet, and Haiku. Whether you are performing high-stakes security audits, routine feature implementation, or simple documentation updates, this skill offers clear escalation rules, thinking mode parameters, and phase-based selection strategies to ensure you get the best results while managing your token budget effectively.

Key Features

Clear escalation and downgrade rules based on task complexity
0 GitHub stars
Token cost optimization strategies to reduce overhead by up to 80%
Task-to-model mapping for every phase of the development lifecycle
Phase-specific recommendations from discovery to final review
Strategic guidelines for enabling and budgeting Thinking mode

Use Cases

Optimizing project budgets by offloading test generation and documentation to Haiku
Determining if a bug fix requires deep reasoning (Opus) or standard implementation (Sonnet)
Deciding the appropriate thinking token budget for complex architectural decisions

About

Key Features

Clear escalation and downgrade rules based on task complexity
0 GitHub stars
Token cost optimization strategies to reduce overhead by up to 80%
Task-to-model mapping for every phase of the development lifecycle
Phase-specific recommendations from discovery to final review
Strategic guidelines for enabling and budgeting Thinking mode

Use Cases

Optimizing project budgets by offloading test generation and documentation to Haiku
Determining if a bug fix requires deep reasoning (Opus) or standard implementation (Sonnet)
Deciding the appropriate thinking token budget for complex architectural decisions