Model-Specific Tips

Different AI models have different strengths. Here's how to get the best results from each.

OpenAI Models (GPT)

GPT-5 / GPT-4o

Strengths:

• Following complex instructions
• Coding and technical tasks
• Structured outputs

Tips:

• Be very explicit about format requirements
• Use system-like instructions at the start
• Request step-by-step reasoning for complex tasks

"Structure your response as follows:
1. Summary (2 sentences)
2. Analysis (bullet points)
3. Recommendation (clear conclusion)

Now, analyze this business proposal: [content]"

o3 (Reasoning Model)

Strengths:

• Complex reasoning
• Math and logic
• Multi-step problems

Tips:

• Don't ask it to explain during reasoning—let it work
• Best for problems with clear right/wrong answers
• May be slower but more accurate

"Solve this logic puzzle: [puzzle]

Take your time to work through it carefully."

Anthropic Models (Claude)

Claude 4 Opus / Sonnet

Strengths:

• Nuanced writing
• Long document analysis
• Following complex guidelines
• Acknowledging uncertainty

Tips:

• Claude responds well to detailed context
• Explicitly state constraints and guidelines
• Ask for Claude's reasoning when useful

"You're helping me write documentation for developers.
Guidelines:
- Use clear, concise language
- Include code examples
- Assume familiarity with JavaScript
- Avoid jargon when possible

Write documentation for this function: [code]"

Claude Haiku

Strengths:

• Speed
• Simple tasks
• Cost efficiency

Tips:

• Keep prompts concise
• Don't expect complex reasoning
• Great for quick lookups and simple transformations

Google Models (Gemini)

Gemini 2.5 Pro

Strengths:

• Multimodal (images + text)
• Factual accuracy
• Research tasks
• Large context window

Tips:

• Great for analyzing images and documents together
• Use for research that needs web information
• Can handle very long documents

"Analyze this image of our website homepage and
compare it to the design specification I'm pasting below.
Note any discrepancies.

[Design spec text]"

Gemini Flash

Strengths:

• Very fast responses
• Good for conversational use
• Efficient for simple tasks

Tips:

• Best for interactive, quick tasks
• Less suited for complex analysis
• Great for brainstorming

Open Source Models

DeepSeek / Qwen

Strengths:

• Strong coding abilities
• Good reasoning
• Cost-effective

Tips:

• Clear, explicit prompts work best
• Good for code-heavy tasks
• May need more guidance than proprietary models

Llama

Strengths:

• Creative tasks
• Conversation
• General knowledge

Tips:

• Works well with conversational prompts
• Good for creative writing
• May be less precise on technical details

Choosing Based on Task

Cross-Model Prompting Tips

What Works Everywhere

• Clear, specific instructions
• Providing relevant context
• Specifying output format
• Breaking complex tasks into steps

What Varies

• Level of detail needed in prompts
• How to request specific formats
• Handling of ambiguity
• Creative interpretation

Adapting Prompts

If a prompt works well with one model but not another:

1. Add more detail - Some models need more explicit instructions
2. Simplify - Some do better with cleaner prompts
3. Change format - Try different ways of structuring
4. Add examples - Few-shot learning helps most models

Next Steps

• Review prompting basics
• Learn advanced techniques
• Browse the prompt library