GPT-5: Your New Favorite AI Tool

This Week

I’ve been glued to GPT-5 since the moment it launched, and I’m loving the upgrades. 

It’s noticeably faster, smart enough to choose the right approach for whatever I throw at it, and the new Vibes feature makes it easy to match its tone to the task. I’ve been testing it on everything from subject lines and reports to critical feedback and brainstorming, and it keeps surprising me.

In this week’s newsletter, I’ll walk you through the biggest improvements, what they mean for your work, and a few fun ways to put GPT-5 to the test.

Let’s get into it.

What You Need to Know About GPT-5

GPT-5, OpenAI’s long-awaited flagship model, dropped last week, and the internet promptly lost its mind.

So, what’s the big deal?

In short, GPT-5 outperforms its predecessors on just about every front. OpenAI is calling it their “smartest, fastest, most useful model yet” and positioning it as a major step toward AGI.

Of course, plenty of people spent the week trying to trip the new model up (like Deedy on X) but when I ran the same tests, GPT-5 nailed the answer every time. Whether it was a math equation or a tricky word problem, it held up.

Beyond accuracy, GPT-5 comes with a set of upgrades that make it faster, more versatile, and easier to use. Here are a few of the ones I found most useful:

You Don’t Need to Pick a Model

Ever stared at ChatGPT, wondering if you should use GPT-4.5 or GPT-3o, and not really knowing the difference? Same.

With GPT-5, you don’t have to worry about picking the right model for the job—it will do that for you.

It can also handle multimodal inputs on its own. Upload a chart, and it’ll use its vision system to extract the data, its reasoning system to interpret it, and its math skills to explain it. In short, GPT-5 can see or hear something, then apply step-by-step logic to make sense of it.

It Admits When It’s Wrong

GPT-5 has been trained to say “I don’t know” when it’s unsure. That’s a big improvement over the overconfident guessing you might’ve seen in models like GPT-4o.

It’s also better at backing up its answers with sources and adjusting the level of detail to match what you actually need.

It Can Handle More Inputs

GPT-5 comes with a much larger context window, able to process entire books, long videos, or complex documents. It supports up to 400,000 tokens (up from 128k in GPT-4o) and no longer loses track of the conversation. That puts it in second place behind Gemini, which currently holds the record with 1 million tokens.

It Matches Your Vibe

Not every conversation calls for the same tone. With its new “vibes” feature, you can set the AI’s style, personality, and emotional approach. This makes it easier to adapt responses to your task. Here are a few “out of the box” options:

• The Listener: Supportive and calm—ideal for customer interactions or coaching conversations.

• The Nerd: Detail-oriented and curious—perfect for research, data analysis, or complex client briefs.

• The Robot: Neutral and precise—best for formal documentation, policy updates, or technical instructions.

• The Cynic: Dry humor and skeptical—great for creative brainstorming, challenging assumptions, or adding personality to campaigns.

Reading about Vibes is one thing, hearing them is another. I tested each one so you can get a feel for how they sound:

The Benchmarks Prove It

On performance tests, this GPT-5  tops the charts. It ranks first across all major categories in the LMArena. It also nailed a perfect score on the AIME 2025 math problems and scored 42% on Humanity’s Last Exam—just shy of Grok 4 Heavy’s top result.

What does that mean? Faster, more accurate answers across a wider range of tasks, whether you’re troubleshooting code, crunching sales data, drafting proposals, or building marketing campaigns.

While GPT-5 is still a long way from true AGI, it’s a clear sign that AI is getting smart enough to leverage itself. You no longer need to understand the nuances between models to get the best results—it figures that out for you.

The base model is already rolling out to free ChatGPT users, with Plus subscribers ($20/month) getting higher usage limits. In other words, you can try it right now.

Missing the Old GPT Models?

If you’ve been lamenting the loss of your favorite ChatGPT model, good news — you can get it back. There’s a hidden setting that lets you resurrect the classics, from GPT-4o to o3 and beyond.

Here’s how to unlock the ChatGPT vault:

• Open ChatGPT and sign in

• Click on your profile > Settings > Toggle on ‘Show Legacy Models‘

• Click on the model selector at the top of the chat (where it says ChatGPT-5)

• Click on Legacy Models from the list

• From the dropdown, you can now select any previous ChatGPT model—including GPT-4o, o3, and older.

You’re welcome.

Watch GPT-5 in Action

Reading about new features is one thing. Watching them in action is better. In this demo, GPT-5 builds a fully working app in real time, showing off its skills in coding, writing, visual perception, and even health-related problem solving.

Watch the video.

Full Summer Update from OpenAI

Still want more GPT-5? You can watch the full 1 hour and 17 minute summer update from the OpenAI team, complete with demos, deep dives, and plenty of behind-the-scenes insights.

Watch it here.

Want to Level Up Your AI Game?

If your team is ready for a hands-on AI strategy session, my custom-designed workshops are built to uncover the workflows that can save you hours every week.

Prefer to start small? My YouTube channel is packed with quick, practical “how-to” videos that show you exactly how I use AI tools for marketing, content, and automation.

Planning an event or conference? I deliver high-energy AI sessions that engage audiences and leave them with actionable strategies they’ll talk about long after the event. Book me for your event here.