I was talking to a friend the other day, and she told me she was obsessed with Perplexity. Why? Because “ChatGPT can’t search the web”.
I was shocked.
ChatGPT actually CAN search the web, and even better than Perplexity.
How come my friend, who works in tech, does not know it?
Because people think they know ChatGPT, but they are talking about a ChatGPT of the past. The one from 2 years ago.
Since then, it can:
Browse the web. Even create 20-page reports with Deep Research.
Store “memories” of the most important things about you.
Remember everything you mentioned before.
Generate quick videos with Sora.
Make images.
So what happened?
People are stuck in the past & their first impression of an app.
Let me refresh your memories with an OLD vs. NEW battle of your favorite AIs.
I covered ChatGPT. Let’s cover some more:
Claude
In the past, Claude was known for:
Having a better writing style than ChatGPT.
Being better at coding.
But you couldn’t search the web.
And it couldn’t perform “deep reasoning”.
But now, Claude can:
Browse the web.
Perform “Extended thinking” (= deep reasoning).
Code extremely well, with a preview of what you coded.
Still, Claude can’t:
Generate images or videos.
Remember your past conversations.
Perform “Deep Search” like ChatGPT to create reports.
PS: Some people say Claude is still better for minor languages, like Hebrew.
Gemini
In the past, Gemini was known for:
Being called “Bard”.
Being laughable because of how bad it was.
Generating the worst images with terrible ethical alignment.
But now, Gemini can:
Browse the web.
Code - and it’s known as the #1 coder (both Gemini-2.5-Flash & Pro).
Generate images with “Imagen 4” and best-of-its-kind videos with Veo-3.
Gemini is the leading model (when you use it through the API)… but still does not compete with ChatGPT when it comes to its chat interface.
Perplexity
In the past, Perplexity was known for:
Being the only one to search the web with AI.
But now, Perplexity…
… is not the only one anymore.
Perplexity is investing in a new browser (Comet) and wants to compete with Google Chrome.
It wants to be the #1 AI browser + AI agent inside your browser.
I canceled my Perplexity subscription, but I’m expecting a lot from them in the future. I believe in their CEO’s vision & roadmap.
Midjourney
In the past, Midjourney was known for:
Being the #1 and only AI to generate images (with Stability & Dall-E).
Messing up your hands when you generate an image.
Only available on Discord.
But now:
Midjourney is still number 1. And that’s incredible.
It has its own website at midjourney.com (no need for Discord anymore).
You do have competitors like Imagen-4 (Google), Dall-E (ChatGPT), Ideogram, Recraft, Flux…
… but Midjourney remains the best for artistic, stunning images.
It still can’t generate good images with text (= infographics or billboards).
Midjourney is quite secretive about its future.
But some rumors say they are going after 3D animation.
Synthesia
In the past, Synthesia was known for:
AI avatars for companies.
But HeyGen would be the one for creators.
But now:
Synthesia is valued at $2.1bn while HeyGen is valued at $500M.
I used Synthesia to create my “creator avatar” & made 60 million views.
I met their teams in London, and I’m quite excited about their future!
Gamma
In the past, Gamma was known for:
The best AI to generate quick slide decks.
But now, Gamma can:
Create an entire website (as a slide deck).
Access every AI model for images (Flux, Imagen…).
Edit your AI images (inside your slides) with written prompts.
And it’s still the best AI to generate quick slide decks.
Opus Clip
In the past, Opus Clip was known for:
You have one long video → you turn it into a TikTok video.
You could also do it from a YouTube link.
But now, Opus Clip has:
A new mode called “Opus Search” for media.
It’s to search through 1,000+ videos, easily. Your whole catalogue.
They also have “Clip Anything” to clip things based on what you see on a video (and not just through the transcribed text of the video). For eg, “Clips of Michael Jordan dunking in this game.”
This is my first Substack & I migrated my previous list of 56,000+ subscribers!
I would love you to support me in 2 ways:
Forward it to your best friend who wants to know “How to AI”.
Do step 1. That’s really just it.
For the newcomers:
It’s my free weekly newsletter on mastering the “How” of “AI”.
My newsletter remains free :) and will always be free!
I want us to master AI before it masters us.
- Yours truly human, Ruben
I’ve been following your work on LinkedIn for quite a while, Ruben, you’ve always been one of the people bringing such sharp, actionable insights to this space! You were actually one of the voices that pushed me to experiment more deeply with prompting, and that journey has now led me to build AI products using LLM APIs, craft GPTs and automations, and dive into AI tools nonstop. That’s actually what I now write about in my own newsletter.
Really happy to see you here on Substack & looking forward to learning even more from your posts here.
P.S. Can't wait to find out about how it was to meet the F1 team!!! Just watching the Monaco race rn :))
AI is nothing like I've ever seen. I'm 69 years old, fascinated with it, and finding ways to use it each day as I work on a new business venture. I've always been a "techie" in my career and personal life, but nothing has moved as fast as AI has, and for this old brain, it's tough sometimes to keep up with it. I've been following you for a while, Ruben, and I always look forward to your newsletters. Keep up the great work!