GPT-5 / GPT-5.1 Guidelines for Tour Operator

Comprehensive guide for using OpenAI GPT-5 series models with the Tour Operator WordPress plugin.

Quick Navigation

Main Documentation

• ← Back to README
• AGENTS.md - AI operations system overview
• Custom Instructions - Copilot configuration

AI Operations Components

• Instructions - Coding standards
• Prompts - Task templates
• Agents - Specialized agents
• Chat Modes - Interactive workflows

Related Guides

• CLAUDE.md - Claude (Anthropic) optimization
• GEMINI.md - Gemini (Google) optimization
• WordPress Packages - Package reference

About This Guide

This document provides GPT-5 series optimization strategies for Tour Operator development. GPT-5 excels at complex reasoning, multi-step planning, and architectural decisions.

Model Comparison

GPT-5.1 (Preview - Cutting Edge)

• Best for: Deep reasoning, multi-step problem solving, architecture-level analysis
• Context window: 128K tokens
• Premium multiplier: 1x
• Capabilities: Agent mode
• Status: Public Preview
• Use cases:
  • Complex architectural decisions
  • Performance optimization
  • Security audits
  • Multi-file refactoring

GPT-5 (GA - Recommended for Deep Reasoning)

• Best for: Complex reasoning, code analysis, technical decision-making
• Context window: 128K tokens
• Premium multiplier: 1x
• Capabilities: Reasoning
• Use cases:
  • Complex multi-step problem solving
  • Complex algorithm design
  • Strategic refactoring plans
  • Performance optimization strategies

GPT-5 mini (GA - Recommended for Speed)

• Best for: Fast, accurate code completions and explanations
• Context window: 128K tokens
• Premium multiplier: 0x (included)
• Capabilities: Agent mode, Reasoning, Vision
• Use cases:
  • Quick bug fixes
  • Simple code generation
  • Inline suggestions

GPT-Optimized Workflows

1. Complex Block Architecture

Use GPT-5 for planning:

Design the architecture for a booking system that includes:
- Tour selection with date picker
- Pricing calculation with seasonal rates
- Multi-step booking form
- Payment integration
- Booking confirmation

Provide:
1. Component breakdown
2. Data flow architecture
3. State management strategy
4. Integration points
5. Testing strategy

Use GPT-5-Codex for implementation:

Based on the architecture, implement the booking-form block:
- Use @wordpress/components
- Integrate with tour pricing API
- Handle validation
- Support theme.json styling
- Include accessibility features

2. Performance Optimization

Use o1 for analysis:

Analyze tour listing performance issues:
- Current load time: 3.2s
- Database queries: 47 per page
- Largest contentful paint: 2.8s

Provide:
1. Performance bottleneck analysis
2. Optimization strategy
3. Implementation priorities
4. Expected improvements

Use GPT-5-Codex for implementation:

Implement caching strategy for tour listings:
- Use WordPress transient API
- Cache taxonomy queries
- Optimize meta queries
- Implement lazy loading

3. Multi-File Refactoring

Use GPT-5-Codex with full context:

Refactor tour booking system across:
- includes/classes/class-tour-booking.php
- includes/classes/class-tour-pricing.php
- includes/classes/class-booking-validation.php
- src/blocks/booking-form/index.js

Goals:
- Consolidate duplicate code
- Improve error handling
- Add comprehensive validation
- Enhance accessibility

GPT-Optimized Prompts

Recommended Prompts

Prompt	Model	Why GPT	Complexity
block-architect	GPT-5	Strategic architecture	Very High
refactor-plan	GPT-5	Multi-step planning	High
performance-audit	GPT-5-Codex	Deep analysis	High
security-audit	GPT-5-Codex	Critical accuracy	High
code-review	GPT-5-Codex	Comprehensive review	Medium
bugfix-risk	GPT-5-Codex	Risk analysis	Medium

View all prompts →

GPT-Optimized Agents

Recommended Agents

Agent	Model	Why GPT	Output
block-architect	GPT-5	Complex architecture	Architecture docs
code-governor	GPT-5-Codex	Quality enforcement	Quality report
block_audit	GPT-5-Codex	Comprehensive audit	Audit report

View all agents →

GPT-Optimized Chat Modes

Recommended Chat Modes

Mode	Model	Why GPT	Workflow
block-plugin-developer	GPT-5-Codex	Complete plugin dev	8-12 turns
refactor	GPT-5	Complex refactoring	6-10 turns
review	GPT-5-Codex	Thorough review	4-6 turns
reviewer	GPT-5-Codex	Deep dive	8-12 turns

View all chat modes →

Best Practices

1. Leverage Deep Reasoning (GPT-5)

For architectural decisions:
1. Explain the problem completely
2. Include constraints and requirements
3. Request step-by-step reasoning
4. Ask for alternatives with trade-offs

2. Use Structured Prompts (GPT-5-Codex)

Provide response in this format:

## Problem Analysis
[Analysis of current state]

## Proposed Solution
[Detailed solution]

## Implementation Steps
1. Step 1
2. Step 2

## Testing Strategy
[How to validate]

## Potential Issues
[Edge cases and risks]

3. Multi-File Context

Include all related files in context:

Context files:
- src/blocks/tour-card/index.js
- src/blocks/tour-card/edit.js
- src/blocks/tour-card/block.json
- includes/blocks/class-tour-card.php

Task: Refactor for performance

4. Iterative Refinement

1. Initial: "Design booking form architecture"
2. Refine: "Add payment gateway integration"
3. Optimize: "Improve validation error handling"
4. Finalize: "Add comprehensive tests"

Task-Specific Guidelines

Block Architecture

Use GPT-5 for planning:

Design a tour search block with:
- Location-based filtering
- Date range picker
- Price range slider
- Tour type taxonomies
- Real-time search results

Requirements:
- Server-side rendering for SEO
- Client-side filtering for UX
- Accessible keyboard navigation
- Mobile-responsive design
- Performance: <100ms filter updates

Provide:
1. Component architecture
2. State management approach
3. API integration strategy
4. Accessibility considerations
5. Performance optimization plan

Security Audit

Use GPT-5-Codex for analysis:

Audit booking system security:
- Payment processing
- User data handling
- Admin capabilities
- AJAX endpoints
- Nonce verification

Check for:
- SQL injection vulnerabilities
- XSS vulnerabilities
- CSRF protection
- Data sanitization
- Authorization checks

Provide:
1. Security issues found (with severity)
2. Exploitation scenarios
3. Fix recommendations
4. Testing checklist

Performance Optimization

Use GPT-5 for strategy:

Optimize tour listing performance:

Current metrics:
- Load time: 3.2s
- TTFB: 800ms
- Database queries: 47
- Memory usage: 125MB

Goal metrics:
- Load time: <1.5s
- TTFB: <200ms
- Database queries: <15
- Memory usage: <64MB

Provide:
1. Performance analysis
2. Optimization strategy
3. Implementation priorities
4. Expected improvements
5. Monitoring approach

Integration with VS Code

GitHub Copilot

# Use GPT-5 mini for inline suggestions
# Copilot automatically uses GPT models for:
- Code completion
- Inline chat
- Quick fixes

# Specify model for complex tasks
@workspace /gpt-5 Refactor this booking system

Copilot Chat

# Architecture planning with GPT-5
@workspace /gpt-5 Design the booking system architecture

# Implementation with GPT-5-Codex  
@workspace /gpt Implement the booking form component

# Quick fixes with Mini
@workspace /gpt-5-mini Fix this validation error

Prompt Engineering Tips

For GPT-5/GPT-5.1 (Reasoning Models)

✅ Do:

• Provide complete context upfront
• Ask for step-by-step reasoning
• Request alternatives and trade-offs
• Include constraints and requirements

❌ Don't:

• Rush the response (deep reasoning takes time)
• Provide partial context
• Ask for quick iterations

For GPT-5-Codex (Code Production Model)

✅ Do:

• Use structured output formats
• Request code with comments
• Ask for complete implementations
• Include test requirements

❌ Don't:

• Assume context from previous unrelated chats
• Request creative variations (use GPT-5)
• Expect deep reasoning (use GPT-5)

For GPT-5 mini (Fast Model)

✅ Do:

• Use for quick bug fixes
• Request simple code snippets
• Ask for straightforward explanations

❌ Don't:

• Use for complex architecture
• Expect comprehensive analysis
• Request multi-file refactoring

Common Workflows

1. New Feature Development

Phase 1: Architecture (GPT-5)
- Design system architecture
- Plan component structure
- Define API contracts

Phase 2: Implementation (GPT-5-Codex)
- Scaffold components
- Implement business logic
- Add error handling

Phase 3: Optimization (GPT-5-Codex)
- Performance tuning
- Security hardening
- Accessibility improvements

Phase 4: Testing (GPT-5 mini)
- Generate unit tests
- Create E2E tests
- Write test documentation

2. Bug Investigation

Step 1: Analysis (GPT-5-Codex)
@workspace /gpt Analyze this bug:
- Error message: [paste error]
- Relevant code: [paste code]
- Steps to reproduce: [list steps]

Step 2: Root Cause (GPT-5 for complex bugs)
@workspace /gpt-5 Determine root cause considering:
- Database state
- User permissions
- Cache interactions
- Third-party integrations

Step 3: Fix Implementation (GPT-5-Codex)
@workspace /gpt Implement fix with:
- Code changes
- Tests
- Documentation

3. Code Review

Step 1: Automated Review (GPT-5-Codex)
@workspace /gpt Review this PR for:
- Code quality
- Security issues
- Performance concerns
- Accessibility compliance

Step 2: Architectural Review (GPT-5 for complex changes)
@workspace /gpt-5 Review architectural implications:
- System design impact
- Scalability concerns
- Maintenance burden
- Migration strategy

Optimization Tips

Faster Responses

• Use GPT-5 mini for simple tasks
• Provide focused context (not entire codebase)
• Use specific prompts
• Request code-only responses when appropriate

Better Accuracy

• Use GPT-5 for complex reasoning
• Use GPT-5-Codex for production code
• Provide complete context
• Include relevant files
• Reference WordPress standards
• Validate with tests

Cost Optimization

• Use GPT-5 mini for development iterations (0x multiplier)
• Use GPT-5-Codex for production code
• Use o1 only for complex planning
• Cache common responses
• Minimize context size

Limitations & Considerations

Model-Specific Limitations

GPT-5.1 (Preview):

• Slower responses (deep reasoning takes time)
• Best for planning, not rapid implementation
• 1x premium multiplier

GPT-5:

• Context limit: 128K tokens
• Best for complex reasoning and debugging
• Doesn't support Agent mode

GPT-5-Codex:

• Optimized for code, may be less creative
• 1x premium multiplier
• Great for complex engineering tasks

GPT-5 mini:

• Less capable with complex tasks
• May produce simpler solutions
• Best for straightforward problems
• 0x multiplier (free!)

Code Quality

• Always review generated code
• Test thoroughly before committing
• Verify WordPress standards compliance
• Check security implications
• Validate accessibility

Resources

• OpenAI Platform
• GPT-5 Documentation
• GitHub Copilot AI Models
• AI Model Comparison
• Tour Operator Docs
• WordPress Block Editor

Comparison Matrix

Feature	GPT-5.1	GPT-5	GPT-5 mini	Gemini 2.5 Pro	Claude Sonnet 4.5
Reasoning	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐⭐
Speed	⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐
Context	128K	128K	128K	1M	200K
Multiplier	1x	1x	0x	1x	1x
Code Quality	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐⭐
Architecture	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐	⭐⭐⭐	⭐⭐⭐⭐
Scaffolding	⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐
Agent Mode	✅	❌	✅	❌	✅
Vision	❌	❌	✅	✅	❌

Use GPT When You Need

✅ Complex Reasoning (GPT-5/5.1)

• Architectural decisions
• Strategic planning
• Algorithm design
• System design

✅ Production Code (GPT-5-Codex)

• Complete implementations
• Security-critical features
• Performance optimization
• Multi-file refactoring

✅ Quick Tasks (GPT-5 mini)

• Bug fixes
• Simple features
• Code formatting
• Documentation updates

Use Gemini When You Need

✅ Speed & Context

• Massive context (1-2M tokens)
• Rapid prototyping
• Batch operations
• Quick scaffolding

Use Claude When You Need

✅ Accuracy & Analysis

• Accessibility audits
• Detailed reviews
• Code quality analysis
• Long-form documentation

Getting Help

Model Selection Confusion?

Start here:

1. Simple task → GPT-5 mini (0x multiplier!)
2. Standard implementation → GPT-5
3. Complex planning → GPT-5.1
4. Code-optimized → GPT-5-Codex
5. Need speed → Gemini 2.5 Pro
6. Need accuracy → Claude Sonnet 4.5

Common Issues

"GPT-5 is too slow" → Use GPT-5 mini for implementation tasks

"Need more complete code" → Request explicit completeness, or use GPT-5-Codex

"Need larger context" → Use Gemini 2.5 Pro (1M tokens) or Gemini 3 Pro (2M tokens)

"Need better accessibility analysis" → Use Claude Sonnet 4.5 or Opus 4.1

"Need faster iterations" → Use Gemini 2.5 Pro or Claude Haiku 4.5

Performance Tips

Maximize Speed:

• Use GPT-5 mini for development (0x multiplier)
• Use GPT-5-Codex for production
• Minimize context size
• Request code-only responses

Maximize Quality:

• Use GPT-5.1 for planning
• Use GPT-5-Codex for implementation
• Provide complete context
• Include test requirements
• Reference WordPress standards

Balance Cost:

• Use GPT-5 mini when possible (0x multiplier)
• Use GPT-5-Codex for most tasks (1x)
• Reserve GPT-5.1 for complex decisions
• Cache common responses