Model-Specific Prompting: Choosing the right AI for your task
In this chapter, we will explore how to strategically choose the right AI model for specific tasks and adapt your prompting approach accordingly.
Model Capabilities: A Practical Comparison
Different AI models have different strengths, weaknesses, and communication styles. Effective prompt engineers adapt their prompts to the model they're using.

Adapting Your Approach By Model
For ChatGPT
Leverages detailed instructions well
Business Application
"Analyze these quarterly sales figures by region. Identify the top 3 underperforming products and suggest potential causes for the decline based on the seasonal patterns in the data."
Personal Use
"Evaluate these 5 investment options for my retirement portfolio based on my risk tolerance (moderate), time horizon (20 years), and current allocation (60% stocks, 30% bonds, 10% cash)."
For Claude
Excels with long documents and nuanced ethical questions
Business Application
"I'm uploading our 50-page employee handbook. Identify policies that might need updating based on current remote work trends and highlight any sections with potentially outdated legal language."
Personal Use
"Review this rental agreement I've pasted below. Flag any concerning clauses and explain what they mean in simple terms. Then suggest points I might negotiate before signing."
For Open Source Models (e.g., LLaMA)
Often require more explicit, structured prompting
Business Application
"FORMAT: JSON output. TASK: Categorize these customer feedback comments into: 'Product Issue', 'Service Complaint', 'Positive Feedback', or 'Feature Request'. COMMENTS: [list of comments]"
Personal Use
"STEPS: 1) Analyze this workout data 2) Calculate average pace 3) Compare to last month 4) Suggest improvements. DATA: [workout statistics]"
Try Prompt Genie to create super prompts for all your AI tasks.
ROI Analysis: Model Selection Case Study
A legal team compared using GPT-5 versus Claude for contract analysis:
GPT-5
Better at identifying subtle contractual risks, but required breaking long contracts into chunks
Claude
35% faster workflow with long contracts due to larger context window
Outcome
22% cost savings by matching the right model to specific contract types
Today's Action
- Identify three different tasks you regularly use AI for
- Test the same prompt across different models if you have access (ChatGPT, Claude, etc.)
- Note differences in response quality, depth, and accuracy
- Create a simple "routing guide" for which model to use for which types of tasks
Advanced Tip
When working with specialized domains (medicine, law, engineering), evaluate model performance on domain-specific knowledge. Some models have stronger performance in certain fields based on their training data.