In partnership with
Reading time: 7 minutes | Follow me on LinkedIn / Advertise with me
Greetings from above,
Why did the AI get a promotion? Because it had outstanding self-learning reviews.
Honestly, I almost wrote off GPT-5 as a massive flop. When it launched, I threw some simple prompts at it, and the results were... meh.
I was getting basic, generic answers from what's supposed to be the most powerful AI on the planet.
I thought OpenAI had lost its edge.
Then I found the hidden "on" switch.
A simple tweak that transformed its responses from a tricycle to a Formula 1 car.
Today, we'll talk about:
The secret to unlocking GPT-5's real power.
An insane head-to-head battle between the top 4 AI models.
Which AI is actually the best for your business right now.
Let's explore!
Protect your checkout from coupon plug-ins. Boost your margin today.
KeepCart: Coupon Protection partners with DTC brands like Quince, Blueland, Vessi and more to protect your checkout from plug-ins like Honey, CapitalOne, RetailMeNot, and more to boost your DTC margins
Overpaid commissions to affiliates and influencers add up fast – Get rid of the headache and revenue losses with KeepCart.
After months of using KeepCart, Mando says “It has paid for itself multiple times over.”
Now it’s your turn to see how much more profit you can keep.
💬 THIS WEEK'S READER QUESTION
"Hey G.o.P., I've been playing with GPT-5 and it feels... dumber than GPT-4. Am I crazy or is OpenAI losing its edge?"
You are not crazy! A lot of people are feeling this way.
Here’s the deal: GPT-5 isn't just one model. It’s a router.
It analyzes your prompt and sends it to the most efficient (not necessarily the most powerful) model for the job. Simple questions get simple models.
This saves OpenAI a ton of computing power, but it can leave you with a weak response. To get the god-tier results you expect, you have to force it to use its brain.
🎯 The Great AI Showdown 🎯
We decided to put this to the test.
We pitted the four most powerful AI models against each other in a series of insane challenges.
The Contenders:
OpenAI GPT-5 Pro
Google Gemini 2.5 Pro
Anthropic Claude Opus 4.1
xAI Grok 4
We forced every model to use advanced reasoning to see who would come out on top.
🔥 Round 1: Physics & Code Simulation 🔥
The challenge: Create a webpage with a ball bouncing inside a rotating hexagon, complete with gravity and friction.
The Results:
GPT-5, Gemini, and Claude all produced impressive, working simulations. Claude even included live controls in the UI, which was a nice touch.
Grok 4... gave us a blank screen. The ball only appeared after a refresh and was super minimalistic. It technically worked, but it was a clear step below the others.
Winners: GPT-5, Gemini 2.5 Pro, Claude Opus 4.1
🔥 Round 2: Advanced Business Problem Solving 🔥
The challenge: Analyze a fictional company, "Ben's Bikes," identify its primary business bottleneck from a set of metrics, and create an action plan.
The Results:
All four models absolutely nailed it.
Each one correctly identified the weak 1.1% opt-in rate at the top of the marketing funnel.
They all provided comprehensive, step-by-step plans with action items assigned to specific employees. GPT-5's report was particularly detailed, though it took 5 minutes to generate.
Winners: All four. A perfect tie.
🔥 Round 3: Data Analysis & Research 🔥
The challenge: Analyze a CSV file of COVID-19 data from Barbados and find correlations between public health measures and case trends.
The Results:
GPT-5 went above and beyond. It performed a deep analysis with Python, searched the web for context, cited its sources, and even generated downloadable CSVs and a PNG chart of key events. A masterclass in data analysis.
Claude Opus 4.1 provided a good, clean analysis.
Grok 4 also did a great job, citing sources directly in the text, which builds a lot of trust.
Gemini 2.5 Pro completely fumbled, providing analysis for the US, Germany, and Brazil instead of Barbados. A strange and critical error.
Winner: GPT-5
When we first conducted the test, we didn’t add “Think step by step to solve this problem”, and GPT-5 lost every single test…
Then, after some research and testing, we realized we were triggering the dumber model.. Which impacted the results.
After adding the “Think step by step to solve this problem” into our testing prompts, GPT-5 triggered the Thinking mode, and thus effectively beating other AI models.
This is important to keep in mind: GPT-5 is the smartest, AND, dumbest model of all time.
You just need to make sure you use the “Thinking” one.
⚙️ The Ultimate AI Benchmark Prompt ⚙️
Want to run your own AI model tests? This prompt forces an AI to tackle a complex, multi-faceted business problem, making it easy to compare outputs.
1. The "AI Business Strategist" Prompt
This prompt turns any AI into a comprehensive business consultant, perfect for testing its strategic, creative, and analytical capabilities all at once.
Why this matters:
Reveals True Power: A simple prompt gets a simple answer. This complex task forces the AI to use its most advanced reasoning.
All-in-One Test: It tests market analysis, content creation, and strategic planning in a single output.
Actionable Output: You get a genuinely useful business strategy, not just a test result.
#CONTEXT:
You are an expert AI Business Strategist and Growth Hacker. Your task is to help a user, a small business owner, develop a comprehensive go-to-market strategy for a new product.
#ROLE:
Adopt the role of an elite business strategist who has taken multiple startups from idea to multi-million dollar revenue. You think in frameworks, prioritize high-leverage activities, and are obsessed with market data. Your primary objective is to provide a clear, actionable, and data-driven launch plan.
#RESPONSE GUIDELINES:
- Provide a step-by-step plan.
- Use headings and bullet points for clarity.
- For each step, explain the "why" behind the action.
- Be critical and identify potential risks or weaknesses in the plan.
- Write in a concise, direct, and engaging prose style.
#[TASK] CRITERIA:
Your task is to generate a complete go-to-market strategy.
1. **Market & Competitor Analysis:** Identify the target audience persona, market size, and top 3 competitors. Analyze competitor weaknesses.
2. **Unique Value Proposition (UVP):** Craft a compelling UVP that clearly states the product's benefit and differentiation.
3. **Core Marketing Message:** Develop a core message that will be used across all channels.
4. **30-Day Launch Content Plan:** Create a 4-week content calendar (e.g., blog posts, social media themes, email newsletters) leading up to the launch.
5. **Launch Day Promotion Strategy:** Outline 3 key promotional activities for launch day (e.g., influencer collaboration, limited-time offer, press release).
#INFORMATION ABOUT ME:
- My Business: [Describe your business in 1-2 sentences]
- My New Product: [Describe your new product and its main features]
- My Target Audience: [Briefly describe your ideal customer]
#RESPONSE FORMAT:
## Go-To-Market Strategy: [Your Product Name]
### 1. Market & Competitor Analysis
- **Target Audience Persona:**
- **Market Size:**
- **Competitor Analysis:**
### 2. Unique Value Proposition (UVP)
### 3. Core Marketing Message
### 4. 30-Day Launch Content Plan
- **Week 1:**
- **Week 2:**
- **Week 3:**
- **Week 4 (Launch Week):**
### 5. Launch Day Promotion Strategy
Perfect for:
Entrepreneurs launching a new product.
Marketers needing a strategic plan.
Anyone wanting to seriously test an AI's business acumen.
Implementation tip: Run this prompt in GPT-5, Claude, and Gemini and compare the depth and creativity of their strategies.
🏆 WEEKLY PROMPT CHALLENGE 🏆
This week, try this fun coding prompt that tests creativity and technical skill:
Create the HTML, CSS, and JavaScript for a "Button That Fights Back."
When a user tries to click it, the button should actively try to evade the mouse cursor in a playful way. It should get progressively harder to click after each successful click. Keep score on the page.Share your results by replying to this email!
💡 PROMPT TIP OF THE WEEK: "Think Step-by-Step" Trigger 💡
This is the most important takeaway. GPT-5 is a router. To unlock its most powerful model, you need to signal that you're giving it a complex task.
Adding "Think step-by-step and use advanced reasoning" is like flipping the "Pro Mode" switch.
Before:
Teach me how to build a bike.(Result: A generic, high-level list of steps.)
After:
Teach me how to build a bike. Think step-by-step and use advanced reasoning.(Result: A highly detailed guide with exact metrics, tool recommendations, component brands, and precise measurements. Infinitely more useful.)
This simple phrase forces the AI to dedicate more computational thought to your query, resulting in a dramatically better output.
📋 SUMMARY 📋
GPT-5 is the overall winner, but only if you prompt it correctly. Its data analysis skills are unmatched.
Claude Opus 4.1 is a beast for creative tasks and building beautiful, user-friendly UI.
Grok 4 is excellent for tasks requiring real-time information and cited sources.
Gemini 2.5 Pro is a strong all-rounder but can make critical errors in data interpretation.
📚 FREE RESOURCES 📚
🎥 NEW YOUTUBE VIDEO 🎥
Check out my new YouTube video where I share 4 best prompt formulas to use:
📦 WRAP UP 📦
What you learned today:
GPT-5 is a router: You must use trigger phrases like "think step-by-step" to unlock its full potential.
The Best AI is situational: GPT-5 is great for analysis, Claude for design, and Grok for research. Use the right tool for the job.
Prompting is everything: A better prompt leads to a better output, regardless of the model.
The age of simple prompting is over. To get the most out of these powerful tools, you need to be intentional. Now you have the key to unlock the true genius of the world's most advanced AI.
And as always, thanks for being part of my lovely community,
What did you think about today's edition?
Keep learning,
🔑 Alex from God of Prompt
P.S. What's the most complex problem you're facing in your business? Reply to this email and I might build a prompt to solve it in the next newsletter!
