How to Do Keyword Research with ChatGPT (the Right Way)

With GPT‑5, OpenAI has basically simplified the whole “which model do I pick?” dilemma down to three core modes. Auto. Fast. Thinking. That’s it. But there’s nuance in how and when you should use each one. And trust me, if you get this right, you’ll save yourself time, frustration, and possibly a few grey hairs.

1. GPT5 Model Selection

Auto

Auto is like the GPS of GPT‑5. You tell it where you want to go, and it decides the best route. If your request is simple, it’ll quietly send it through Fast. If it senses complexity, it’ll switch to Thinking without you lifting a finger. It’s smart. But sometimes it’ll overthink a “what’s the capital of Chile?” type of question and route you to Thinking. Slightly annoying, but that’s just how OpenAI optimised GPT‑5 to balance compute efficiency.

OpenAI has been pretty upfront about this: one of the big pushes with GPT‑5 was to cut down on server strain while giving us better reasoning capacity at scale.

Fast

This is your go‑to for 99% of tasks. Quick answers, brainstorming, simple copy. If you’re drafting an email subject line or asking it to summarise a short text, Fast is perfect. It’s built to be responsive – you’ll actually feel the difference. In many ways, this is a little bit what GPT‑4o felt like: snappy, fast responses that made everyday tasks flow smoother... except now its smarter.

Thinking

Thinking mode is where GPT‑5 flexes. It’s slower, yes, but with a massive 196k token context window (about 4x GPT‑4o’s). Use it when you’re doing deep strategy work, coding problems, or anything with multiple moving parts. The trade‑off is speed, but the payoff is accuracy and depth.

Legacy Models

Missing your old friends GPT‑4o, 4.1, o3? They’re still tucked away. Go into Settings → General → Show additional models. Flip that toggle, and you’ll see them under the “Legacy Models” dropdown. Not essential for most people, but nice to have in case you want to compare outputs.

2. Tools You Actually Need to Know

Here’s where GPT‑5 goes from “chatbot” to “Swiss army knife.” These tools are what make the Plus plan worth it.

Add Photos & Files

Upload up to ~20 files at once. Docs, spreadsheets, PDFs, even images. This turns GPT into a multi‑modal assistant, it can read your PDFs, analyse your data, or even look at a photo and give insights. Super useful for research and SEO workflows (think uploading a CSV of keywords or a competitor’s brochure).

Search

Search gives GPT real‑time internet access. Officially, it’s powered by Bing. Unofficially, some SEOs testing the responses claim it’s pulling Google results more often than not. Either way, the point is: this is your way to cut hallucinations and get fresh data.

*Important side note* more and more sites are blocking AI crawlers (Cloudflare being the biggest culprit). So don’t assume GPT can access every corner of the internet. But when it works, it’s gold.

Deep Research

This is not just “search but slower.” It’s a Plus‑exclusive mode that basically turns GPT into your AI research assistant. Instead of a quick lookup, it actually performs multi‑step searches, not just scanning the top answers, but digging deeper into related areas, following interesting leads, and going further down the rabbit hole until it collects all the useful information. It’ll then spend 2–30 minutes combing sources, asking you clarifying questions, and producing a fully cited report.

Limits: ~25 deep research sessions/month on Plus. But that’s plenty if you save it for the big stuff – competitor analysis, market research, or building the foundation for a new content campaign.

Agent Mode

Think of Agent Mode as hiring an AI intern who also knows how to use a computer. It can:

• Browse sites like a human (click links, fill out forms).
• Run code in a terminal.
• Pull data into spreadsheets.
• Generate slide decks with charts.
• Even generate leads.

The key difference: unlike normal GPT where you guide every step, an Agent executes multi‑step workflows autonomously. You stay in control (it’ll ask permission before big actions), but this is the closest thing to an AI employee. And just to be clear, OpenAI isn’t the only one with this kind of autonomous agent tools like Manus or GenSpark offer similar functionality. Personally, I like having everything under one subscription, but it could be worth testing those out too since they bundle in other useful tools.

Canvas

Canvas is my personal favourite. It’s basically a Google Docs‑meets‑GPT editor where you and the AI can write side‑by‑side. Instead of GPT spitting out text in chat for you to copy, it edits inline, with suggestions, highlights, and even “track changes” style edits.

It’s a game‑changer for:

• Blog drafts
• SEO content refinement
• Code debugging
• Client deliverables
• Creating simple but useful HTML apps you can embed in your site

You can even throw a meeting transcript in there and have it convert it into a structured, sharable HTML report.

Create an Image

Built on OpenAI’s GPT‑4 image model (a huge upgrade from DALL·E). It handles text in images very well which used to be the Achilles heel of AI art tools. Great for social graphics, blog headers, or quick visuals when you don’t want to fire up Photoshop. That said, whilst this is an incredible AI image tool, in my opinion it’s by far not the best. If you want images that are almost indistinguishable from real photos, you should check out Flux from Black Forest Labs or Google’s new Nano Banana generator. At the end of this section, I’ll drop three images below from three different image generators with the same prompt: GPT‑4o, Flux, and Nano Banana. This way, you can make up your mind about which one you like better.

Prompt: High-angle over-the-shoulder photo of a person, showing the back of their head and shoulder. Their smartphone is in focus, displaying an Instagram post on the screen with the words "AI image generation." The background is softly blurred to emphasize the phone.

‍

Study & Learn

This is more niche. Essentially GPT builds custom study programs. Perfect if you want to learn a new skill (SEO, coding, algebra homework for your kid). If your kid asks for help with algebra or quadratic equations, stuff you’ve totally forgotten, you can definitely use this tool to get you out of the pickle. Not something every business owner will use daily, but it’s there...

3. Customisation & Settings You Shouldn’t Skip

Here’s where you turn GPT from “a tool” into your tool.

• Custom GPT profile: Add your name, role, and tone. More importantly, use the “traits” field to set shortcuts. Example: prefix with C for concise one‑liners, L for long‑form detailed answers. Total time‑saver.
• Memory: Toggle this on. GPT will remember how you like things done, your writing style, even recurring workflows. You can edit or delete memories anytime. I think this is really the kind of unsung hero of all the GPT settings. The more memory and things that GPT remembers about you, the more personalised your ChatGPT experience will be. Yes, it can feel a little scary having an AI remember so much, but the deeper you go the better it gets. Think of it like Apple’s ecosystem effect: once your iPhone, AirPods and MacBook all work seamlessly together, it’s tough to leave. Same here, the more GPT remembers, the harder it is to imagine switching to another app.
• Connectors: Gmail, Google Drive, Calendar, Notion, Canva. Once hooked up, GPT can fetch info and act directly in those apps.
• Personality features: GPT‑5 also lets you set a personality style. You can make it more formal, friendly, cynical, or even playful depending on what suits you best. This goes beyond shortcuts – it actively changes how the model frames responses, almost like giving it a new voice. For example, a ‘cynic’ personality will give you blunt, witty replies, while a ‘coach’ might be more encouraging and structured. This is brand new in GPT‑5 and worth experimenting with. If you want to go deeper into this, we've created a detailed guide on GPT-5's new personality, which you can check out here.
• Data control: Disable “Improve the model for everyone.” Otherwise, you’re giving OpenAI consent to train on your data. Especially critical if you’re in law, healthcare, or finance.
• Voices & Themes: More cosmetic, but advanced voice mode is actually fun if you like talking to your assistant instead of typing.

4. Best Practices (a.k.a. Don’t Do Dumb Stuff)

• Use Fast by default. Only switch to Thinking when you’re tackling strategy or complex analysis.
• Don’t waste Deep Research. Save it for competitive research or market deep dives, not “what’s the best pizza topping.”
• Lean on Canvas. If you’re drafting or editing, don’t settle for chat bubbles – put it in Canvas and collaborate properly.
• Treat Agent Mode like a junior hire. Give it clear tasks, review the outputs, and never let it run wild with sensitive stuff.
• Stay organised with Projects. For ongoing campaigns, group your chats, files, and settings in one place. Keeps you sane.

‍

FAQ

1. What’s the difference between Fast and Thinking modes in GPT-5?
Fast gives you quick, snappy answers (like GPT-4o), while Thinking takes its time with deeper reasoning and a much larger context window.

2. Can I still use older GPT models like GPT-4?
Yes. Toggle Show additional models in Settings and you’ll see legacy models like GPT-4o, 4.1, and mini versions in the dropdown.

3. How is Deep Research different from normal Search?
Search pulls quick results from the web. Deep Research runs multi-step queries, follows leads, and produces a fully cited report — perfect for competitor or market analysis.

4. What can Agent Mode actually do for my business?
It can browse websites, fill out forms, run code, generate reports, or even schedule meetings — essentially acting like a junior AI employee under your supervision.

5. Is Canvas just for writing?
Not at all. It’s for writing and coding. You can refine SEO drafts, debug scripts, or even build simple HTML apps you can embed on your site.

‍

GPT‑5 arrived in August 2025, and while everyone was busy drooling over bigger context windows and slicker multimodal reasoning, a quieter but far more human feature slipped in: personalities. Yes, your chatbot can now choose a mood. Or, more accurately, you can pick which mood you want it to wear.

This update isn’t a gimmick. Personalities change the way GPT‑5 speaks, reasons, and interacts. No more copy‑paste politeness or “always‑sunny” tone. Instead, you can pick something that matches your task, or your tolerance for small talk. In my own experience, Cynic has quickly become my favorite. It’s blunt, sarcastic, and will tell you what you need to hear without buttering you up. That small twist makes using GPT far less sterile and much more productive.

What Are GPT‑5 Personalities?

GPT‑5 offers four presets, Cynic, Robot, Listener, and Nerd. Each is more than a coat of paint. They shape how the AI chooses words, structures replies, and balances empathy with precision.

• Cynic: Dry humor, snarky edges, practical over polite.
• Robot: Stripped of emotion, factual and to the point.
• Listener: Empathic, reflective, almost therapeutic.
• Nerd: Obsessed with detail, loves examples, enthusiastic.

They’re part of a broader personalization push that also includes memory, user bios, and even custom UI themes. This isn’t just cosmetic. It’s OpenAI acknowledging that style matters as much as accuracy.

‍

How to Set GPT‑5 Personalities

Switching personalities is thankfully easy. On desktop or mobile:

1. Open ChatGPT and click your profile icon.
2. Select Customize ChatGPT.
3. Choose your Personality from the dropdown (Cynic, Robot, Listener, Nerd).

Alternatively, you can pick one at the start of a conversation by clicking the sparkle icon next to the model name. Keep in mind that personalities apply only to new chats. If you start with Cynic, that chat will stay Cynic forever unless you reset it. You can always revert to the Default personality.

Which Personality Should You Use?

Different personalities shine in different scenarios:

PersonalityBest ForCynicCutting through fluff, reality checks, quick decisions. Think strategy sessions or when you’re sick of AI pleasantries.RobotTechnical work, debugging code, analyzing spreadsheets, data crunching. Anything where tone is a distraction.ListenerDrafting sensitive emails, brainstorming during stressful moments, or tackling anything emotional. It reflects and affirms instead of lecturing.NerdDeep dives, research, learning, planning. Ideal for exploring complex topics or asking for structured, example‑rich output.

In my workflow, Cynic wins hands down. It’s refreshing to get advice without sugarcoating. The snark is a bonus, it feels like a colleague who can be both helpful and slightly amused at my mistakes.

‍

Pro Tips & Best Practices

Personalities don’t override GPT‑5’s core abilities. The reasoning, multimodal input, and massive context window still matter more for heavy tasks. But the way the model communicates changes dramatically.

• Match the tone to the task. Use Robot for crunching numbers, Listener for writing HR emails, Nerd for research, Cynic for strategy.
• Expect updates. OpenAI has already promised to make GPT‑5 “warmer” without losing its professional edge. Translation: Cynic may get even sharper, Listener may get even softer.
• Combine with UI personalization. Memory, user bios, and custom colors make personalities feel more integrated. The AI stops being a generic box and becomes a tool that fits you.
• For developers. If you’re using the API, you can mimic personalities by adjusting parameters like verbosity and reasoning_effort to steer tone and depth.

My Experience With Cynic

I’ll admit it: Cynic hooked me fast. The first time it told me, flat out, that my idea was garbage, I almost laughed. Not because it was wrong, it wasn’t, but because finally, an AI wasn’t scared to tell me the truth. No hedging, no over‑apologizing. Just honesty with a hint of bite.

Cynic is particularly good for brainstorming and critical review. When you’re too close to your own work, it feels like having a brutally honest colleague. It doesn’t care about your feelings, which, ironically, makes it far more useful than an overly polite assistant. The snark makes it memorable, too. That unexpected change in tone can jolt you into looking at your work differently.

Conclusion

GPT‑5 isn’t just another “faster, smarter” upgrade. With personalities, it finally gives users control over style, not just content. Choosing the right personality can sharpen tasks, soften conversations, or simply make the AI more enjoyable to use. And if you’re like me, you’ll quickly find that Cynic, sharp, honest, and a little cheeky, becomes your daily driver.

Personalization isn’t optional anymore. You can ignore it and stick with Default, but then you’ll miss out on what makes GPT‑5 actually feel different: an assistant that works not only with your tasks, but also with your temperament.

‍

Frequently Asked Questions About GPT‑5 Personalities

1. Can I switch personalities in the middle of a chat?
No. Personalities are set when you start a new conversation. If you want to change, you’ll need to start a fresh chat with the new personality.

2. Do personalities affect the accuracy of GPT‑5’s answers?
Not really. The underlying reasoning and multimodal capabilities stay the same. What changes is tone, style, and the way responses are structured.

3. Are personalities available for free users?
At launch, they’re mainly available for paid tiers like Plus, Pro, and Team. Free users may see limited or delayed access depending on OpenAI’s rollout.

4. Can I create my own personality beyond the four presets?
Yes. You can type in a custom style description like “casual and witty” or “formal and concise,” and GPT‑5 will adapt its voice accordingly.

5. Which personality is best for professional settings?
It depends on the context. Robot is safest for technical work and reports, Listener is strong for team communication and HR‑related writing, Nerd is perfect for research, and Cynic—if your coworkers can handle blunt honesty—can be great for strategy and critique.

‍

‍

So GPT-5 is here, and as usual, the internet has split into two camps: people who love it, and people who think it’s the worst thing to happen since autocorrect decided to “fix” your texts. Honestly, I think a lot of the hate comes from people not using it correctly. I’ve just spent seven days reading, watching, and pushing GPT-5 to its limits, so here’s the straight-to-the-point guide you actually need.

1. Unlock All the Hidden Models

When you open the model selector, you’ll probably only see:

• Auto
• Fast
• Thinking
• Pro (if you’re on the $200/month plan)

If you scroll down to "Legacy models," you might spot GPT-4o. But here’s the thing: there’s way more available.

Here’s how to unlock them:

• Go to Settings
• Open the General tab
• Toggle Show additional models

Now, under GPT-5, you’ll see Thinking Mini, which is fantastic for copywriting and lighter creative work.

Which one should you use?

• Auto: GPT will pick based on complexity. Simple = Fast, complex = Thinking.
• Thinking: Better for complex, multi-step reasoning.
• Fast: Cheap, efficient, and surprisingly good. Great default.
• Thinking Mini: Like Thinking, but lighter and faster.

OpenAI’s routing system tends to pick the cheaper (Fast) option if possible. You can always start a chat with Thinking for context, then switch to Fast to save time and credits.

And yes, if you’re still emotionally attached to old favorites like GPT-4.1 or GPT-3.5, they live in the Legacy models section. OpenAI brought them back after a very vocal backlash.

2. Play With the New Personalities

GPT-5 now lets you pick a built-in personality. At first, I thought these were a bit of a gimmick, but the more I use them, the more I realize they’re surprisingly useful:

• Cynic: Snarky, blunt, and great for no-nonsense feedback. My personal favorite.
• Robot: Zero small talk. Just gets to the point. Perfect for coding or rapid drafting.
• Listener: Supportive and empathetic. Weird to say about an AI, but some people genuinely use it for emotional support.
• Nerd: Excited, thorough, and detail-obsessed. Perfect for learning something in depth.

Switch them up and see which fits your workflow. I’ve found Cynic is great for constructive criticism, and Robot is a huge time saver.

You can access all these personalities by going to the Customize ChatGPT.

3. Use the Canvas Feature for Apps and Dashboards

Canvas is GPT-5’s built-in interactive workspace. You can:

• Build games
• Create data dashboards
• Make interactive reports

You can share these with anyone, even if they don’t have a ChatGPT account. This makes it a goldmine for lead generation, imagine sending an interactive report that doubles as a pitch.

4. (Gimmick Alert) Change Your Highlight Colour

It’s cosmetic, not functional, but you can change your GPT interface accent colour to blue, green, yellow, pink, orange, or purple. It’s the kind of feature Apple would hype up as a major release, even though it won’t make GPT-5 any smarter. Still, it does make the workspace feel less sterile.

5. Use the Prompt Enhancer to Fix Bad Prompts

If GPT-5 isn’t giving you what you want, stop blaming the model. Prompting is just communication, and bad prompts = bad results. OpenAI’s Prompt Enhancer (available in the Playground) takes your vague or poorly written prompt and turns it into something far more specific.

Poorly defined prompts are one of the biggest reasons people fail to get good AI outputs. In fact, research shows that improving prompt clarity can boost task accuracy by up to 30% (Stanford HAI). Use the enhancer and you’ll instantly see better results.

Bottom line: GPT-5 is incredibly capable, but only if you unlock all the tools and actually use them. Stop treating it like a magic trick and start treating it like the power tool it is.

‍