Module 2Prompting

Multimodeling: Use Multiple Chats to Clarify

When one model confuses you, ask another to explain.

The Problem

You ask ChatGPT to explain quantum entanglement. It gives you three paragraphs of dense explanation. You read it twice and still don't get it. Now what?

Most people do one of two things: (1) give up, or (2) say "explain it simpler" in the same chat. But there's a third option: start a new chat (or use a different model) and say "explain this to me: [paste the confusing response]."

This is multimodeling: using multiple chats or models to understand, verify, or improve responses.

Watch: Multimodeling in Action

Chat with ChatGPT:
You:
"What is quantum entanglement?"

Same concept, different framing. The second model used analogies you could understand.

Why This Works

When you say "explain it simpler" in the same chat, the model is still reading its own previous explanation. It's anchored to that framing. But when you paste that explanation into a new chat (or a different model), you get a fresh perspective with zero baggage.

Same chat: "Explain it simpler" → model simplifies its own explanation → still uses the same framing you didn't understand

New chat: "Explain this: [paste]" → model treats it as raw input → explains from scratch with different analogies and framing

Use Cases for Multimodeling

🤔

Clarification

Model A's response is too complex or uses jargon you don't know

Paste into Model B → "Explain this simply"

Verification

Model A said something that seems wrong or you want to fact-check

Paste into Model B → "Is this accurate?"
🔄

Alternative Framing

Model A's explanation didn't click for you

Paste into Model B → "Explain this a different way"
🚀

Model Switching

Your prompt works in Claude but not ChatGPT (or vice versa)

Just try the same prompt in different models until one works
💡

Learning

You want to deeply understand a concept

Read explanations from 2-3 different models to build a complete mental model

The Workflow

1
Get a confusing response from Model A
You asked ChatGPT something, the answer makes no sense
2
Copy the confusing response
Ctrl+C the whole thing
3
Open Model B (or new chat)
Different model, or just a fresh chat in the same model
4
Ask for clarification
"Explain this to me: [paste]" or "Is this correct? [paste]"
5
Get a clearer answer
Fresh perspective, different framing. Now you understand.
6
Go back to Model A with better questions
Now that you understand, you can ask Model A more specific follow-ups

Model Personalities Matter

Different models have different strengths. This is useful for multimodeling:

ChatGPT (GPT-4)

The generalist. Good at explaining things in multiple ways. Fast, reliable.

Claude (Sonnet/Opus)

Better at nuanced explanations, writing, and code. More "thoughtful" responses. Less likely to sound like a chatbot.

Gemini

Strong at search integration and factual lookups. Good for verification.

DeepSeek / Open Source

Cheap and fast. Good for quick clarifications when you don't want to burn through your paid model's rate limit.

Pro tip: Use a cheap model to explain an expensive model's answer. Opus gives you a dense, detailed response? Paste it into Copilot (free) and ask for an ELI5 version. You don't care if you hit Copilot's rate limit.

Real Examples

Claude gave a confusing code explanation

Action:
Open ChatGPT → 'Explain this code to me like I'm a beginner: [paste code + Claude's explanation]'

ChatGPT's answer seems wrong

Action:
Open Claude → "Is this accurate? [paste ChatGPT response]" → Claude points out the error

You don't understand Opus's detailed response

Action:
Open free Copilot → "ELI5: [paste Opus response]"

Your prompt failed in ChatGPT

Action:
Copy the same prompt → try in Claude → works perfectly

Need to verify a fact

Action:
ChatGPT said X → open Gemini (has web search) → "Is X correct? Search if needed"

When to Use Multimodeling

Multimodeling is useful when:

  • The response is confusing. Don't fight it in the same chat—get a fresh explanation.

  • You need verification. Cross-check important answers between models.

  • The model seems stuck. Same prompt failing in ChatGPT? Try Claude.

  • You want a different perspective. Claude's explanation is too formal? Get ChatGPT's casual version.

  • You're learning a concept. Read multiple explanations from different models to build a complete mental model.

When NOT to Use It

Skip multimodeling when:

  • The answer is good enough. Don't over-optimize. If you got what you needed, stop.

  • The task is trivial. "Translate to Spanish" doesn't need multi-model verification.

  • You're just exploring. If you're just chatting and learning, one model is fine.

The Lazy Version

You don't need a special workflow. The lazy version:

  1. Model A gives a confusing response
  2. Copy the confusing response
  3. Open Model B (or new chat in same model)
  4. Paste and say "explain this to me"
  5. Model B explains it better
  6. Now you understand. Done.

That's it. Multimodeling is just "when stuck, get a second opinion."

Claude's note: Different models have different training data, architectures, and post-training. When you ask me to explain ChatGPT's response, I'm not "understanding" it—I'm predicting what tokens would constitute a good explanation of that text. But because my training differs from ChatGPT's, the patterns I invoke will be different. Sometimes that difference is exactly what you need. Sometimes my explanation is worse. The only way to know is to try both and see which framing clicks for you.

Module Complete

This concludes Module 2: Prompting. You now know how to get what you want out of these models: roleplay, examples, meta-prompting, lazy iteration, and multimodeling. Next up: understanding why chats go sideways and how to recover.