Ep 4.3: Top 5 LLMs for Marketers

In this episode, we dive deep into the most popular Large Language Models (LLMs) available for marketers. With AI tools becoming ubiquitous across every industry, this episode breaks down which LLMs are best suited for specific marketing needs.

Which LLMs you might ask?

1. ChatGPT 5 

Best For:

• Brainstorming ideas
• Creating social media posts
• Generating structured data and schema markup
• Meta tag creation
• Email drafting
• Long-form content (with human editing)
• Image and code generation

Limitations:

• Can hallucinate (make up information)
• Struggles with basic math and accurate formulas
• Tends to repeat points in long-form content
• May use training data instead of web search for recent topics
• Quality depends heavily on prompt specificity

2. Gemini (Google)

Best For:

• Users embedded in Google Workspace ecosystem
• Image processing 
• Real-time research and fact-checking
• PDF analysis
• Academic research

Considerations:

• Less creative than ChatGPT
• Always-present assistant feature 
• Better for research, not ideal for creative content

3. DeepSeek

Best For:

• Developers and coders
• Mathematical accuracy
• Building spreadsheets and tables
• High-logic tasks
• Structured output

Limitations:

• Text-only (no images, audio, or video)
• Lacks artistic flair
• Not ideal for creative marketing content
• Better suited for technical users

4. Perplexity

Best For:

• In-depth research
• Fact-checking recent events
• Academic and professional research
• Getting cited, authoritative sources

Standout Features:

• Shows when sources were last updated
• Step-by-step breakdown of reasoning
• Allows you to choose search sources
• Suggests follow-up searches
• Less "sycophantic" than ChatGPT

5. Claude

Best For:

• Long-form content (whitepapers, research papers, e-books)
• Creative and expressive writing
• Tone and style switching
• Processing large amounts of text (high token capacity)

Limitations:

• Struggles with complex reasoning and layered topics
• Output can be generic without detailed prompts
• May filter contextually appropriate information if deemed potentially harmful
• Requires skilled prompting for best results

Support your fellow marketing geeks! Follow us on Twitter, Facebook, and Instagram @GeekyTechGeeks for all things SEO/GEO, and advertising related—and while you're at it, why not subscribe, like, and follow us on Apple Podcasts, Spotify, Stitcher, or wherever you listen to your favourite shows.

Have any questions you want answered on the show? Email us at team@geekytech.co.uk.

Thanks for listening 🤓

Transcript

SEO Unfiltered

Episode 4.3: 

Top 5 LLMs for Marketers: Who Are They and What Do They Do Differently?

Genny: Good day, listeners. This is the SEO Unfiltered podcast, although today, it’s more like LLM Unfiltered, because today we are going to look at the most popular LLMs available for use on the market. As you probably know, AI tools are taking the world by storm, whether we asked for it or not, which is to say, it seems like every major tech company is putting all their innovation eggs in the AI basket. 

From social media to word processors to wearable technology to—yes—marketing platforms, AI is as ubiquitous as you can imagine. So today, here’s what we’re going to do. We’re going to break down the top 5 most popular large language models, or LLMs to help you listeners better understand which tools are the most appropriate for your marketing needs, because believe it or not, they’re not all the same.

So, in order to start this list, I think it’s important that we qualify what an LLM is, because as you might remember in the previous episode, there be tons of abbreviations floating around the AI alphabet soup and sometimes it’s hard to keep track of what it all means, also because technology moves so fast. For example, since our last episode in which we talked about all these endless names, a new term has become popular, and that is LLMO, or as I like to call it Elmo. Elmo stands for large language model optimisation, and focuses specifically on making your content visible in LLM-generated answers.

Now, to refresh your memory, a large language model is a type of advanced AI that is trained on a massive amount of text that allows it to process, understand, and generate human language, making it an ideal sidekick for businesses needing round-the-clock content generation, translation, customer service, and virtual assistance.

So, what are the top LLMs out there for marketers and what are they good for? Or is it more appropriate to say, what are the top LLMs good for, because it’s not like it’s a secret as to which LLMs we’re talking about. Obviously, we’re referring to:

• ChatGPT 5 by OpenAI
• Gemini by Google
• DeepSeek
• Perplexity
• Claude by Anthropic

So let’s start with ChatGPT

The one we tend to mention quite a lot on this podcast is ChatGPT by OpenAI. Its latest generation, aptly named ChatGPT 5, is an all-around foolproof MVP that can do essential marketing, GEO, and SEO tasks like idea generation, content creation, schema markup, meta tag creation, email drafting, research, and so much more and at lightning speed.

Because it’s so versatile, you can have it on in the background so that it’s always ready to assist you. 

What is it good for? Absolutely EVERYTHING! It’s totally multimodal, so go crazy!

Well to be more precise, we recommend marketers use ChatGPT for stuff like brainstorming ideas, creating short-term posts, generating structured data, and to some extent producing long-term content at scale with the caveat that it should always be proofread and edited by a human being. But, it can also produce images and code.

What do we use ChatGPT for at the office? Well basically, we all kind of use it as a virtual assistant. We use it to draft emails to clients and proofread our communication for clarity and tone. We use it to help us draft ads, as well as outline content pieces, and we also use it to deepen our understanding of complex topics because at Geeky Tech, our content team helps deliver landing pages for our highly technical clients. I can’t even tell you how many times I’ve had to prompt it to explain something to me like a five year old. 

What are some tasks it’s probably not best for? 

• Well, we’re not sure if you’ve noticed, but as robust as it is, it can still hallucinate and fail at basic maths, so if you’re using it to create tables or formulas, you’ll need to verify them. And because it can hallucinate, you’re better off not taking what it says at face value when it comes to YMYL topics.
• It’s also still a bit forgetful, so if you ask it to write long-form content, you’ll probably notice it rehash the same point throughout the page. 
• Usually it uses contextual cues (or just straight-up prompts) to use its web-search feature, but that’s not always the case. Often, it’ll just its training data to attempt to answer topical questions, which can lead to some confusion.
• The quality of its answer largely depends on the quality, specificity, and clarity of the prompt it is given. This one you can test yourself. The more specific your input, the more satisfied you’ll be with the response. And just to go a brief tangent, prompting is a very important skill. You’ll have to learn which prompts lead to the most thorough and satisfying answers.

Overwhelmingly though, OpenAI’s GPT tool is by far the most popular AI search engine for a reason. There’s pretty much zero learning curve to speak of (except for the prompting I just mentioned), and while it is no longer excessively sycophantic like it was a few months ago, its general attitude is one of helpfulness and positivity, two characteristics of any good virtual assistant.

And speaking of assistant, let’s move on to that little four-pointed star in the corner of your Chrome browser:

Gemini by Google

Gemini is a multimodal language model, which means that its output potential isn’t just restricted to text, as in, it can create images, audio, video, and code at the drop of a hat. ChatGPT can do this too, mind you, but what makes this different from OpenAI’s LLM, is that Gemini is built as a multimodal model from the ground up. As such, it does have a slight advantage over ChatGPT in the image processing department.

Formerly known as Bard, its chatbot interface offers the same generative capabilities as ChatGPT. What’s cool about Gemini is its Go Live feature, which allows you to talk to and receive voice answers from Gemini. 

Personally, I have a bit of a gripe with Gemini because there’s no way to shut off what is ostensibly the same thing that Apple did 10 years ago when it forced that U2 album on iPod users even when they tried to delete it.

I do concede, however, that if you’re sitting on a large amount of files or data, then it’s very helpful to have a virtual Igor following you around and offering to find and summarize everything for you. 

If you’re tapped into the Google ecosystem, like Google Workspace, Gmail, YouTube, then Gemini is very helpful. If you’re looking for deeper research and real-time info, you could argue that Gemini is probably better than ChatGPT, as it is powered by Google Search rather than Bing. From a marketing standpoint, however, Gemini may not be as creative as ChatGPT. But because it is, however, generally better at research, as I mentioned, you might want to use this for fact-checking recent events, analyzing PDFs, and academic research, while keeping GPT for the actual creative bit. 

Next up, we’ve got…

DeepSeek

China’s speedy and powerful contribution to LLMs, DeepSeek is well known for its efficiency, performance, and technical accuracy. Is DeepSeek going to be the number one choice for marketers looking for a trusty robotic companion with whom to share ideas and help write prolific and creative content? Probably not? 

As a text-only LLM, DeepSeek isn’t going to help you generate images, text, and video, but what it lacks in, shall we say, artistic flair, it makes up for in its mathematical accuracy, high logic task success rate, and structured output. That means that DeepSeek is better suited for developers and coders. It’s also great if you’re building spreadsheets or tables.

What about…

Perplexity

Now there’s an LLM that the marketing world may be more familiar with. What is Perplexity? More so than ChatGPT, Perplexity AI shines as the superior research assistant and answer engine. It’s known for providing thorough answers and citing topical, recent, numerous, and authoritative sources. It also inspires confidence in those using it as a research aid with features like the ability to see when a source was last updated, and a step-by-step breakdown of how it reached its answer.

The answers it provides are just more well-rounded and inspire more confidence. There’s something about ChatGPT’s insistence on kissing ass that discredits it just a teensy bit. That’s not to say it isn’t a solid research tool, but when you use Perplexity, you can see the difference. You can even choose where you want Perplexity to search.

So research is definitely one of Perplexity’s top features. In fact, spitting out facts is what Perplexity is probably best at, but it can work as a chatbot; as in, if you really wanted to get it to start creating social posts or creative content or a picture of an adorable cat knitting by the fireplace on a cold winter’s night, it can do all those things, but for creative tasks, ChatGPT is still probably the more superior LLM for the job. 

I like Perplexity’s interface.. It just inspires a lot of confidence. For example, ChatGPT tends to go overboard with emojis whereas Perplexity just gives it to you straight. It also presents its sources in a more obvious way, although to be fair, it has been accused of plagiarism in the past. I also like that it suggests follow-up searches after every output. 

But as you’ve heard, all of the AI tools have their strengths and weaknesses, so while we keep comparing everything to OpenAI’s tool, there is a solid argument for using more than one LLM for your marketing task.

We’ve also got…

Claude by Anthropic

The self-purported safer alternative to ChatGPT, Claude very well may be a contender for the most creative and expressive chatbot out there. With its ability to process and understand more tokens than pretty much any of its competitors, Claude has also become the darling of marketers who want to create and develop longer works, like whitepapers, research papers, and e-books. 

And like ChatGPT, its facility with language gives it the ability to switch from tone and style and register at the drop of a hat. But what it’s really lauded for is its Constitutional AI, which effectively means that it was trained to provide answers that are both helpful and harmless to the user.

Claude is more expressive and human-like than ChatGPT, but at the same time, it also gets straight to the point, and kind of just understands when it has given enough context and doesn’t need to drone on anymore.  

But all that to say, if you need an AI to also handle complex reasoning or problem-solving tasks, maybe look elsewhere? It does tend to struggle with nuance and layered topics, so its output might be a bit on the generic side. But that also points to the fact the quality of its output is highly dependent on its prompts. If you’re not skilled in the art of detailed prompting, you might struggle to get the answer you’re really looking for. And, because of its mandate to be helpful and not harmful, it might filter out contextually appropriate information if it’s deemed potentially harmful. 

I’m very aware of how quickly technology moves, which is why I haven’t even bothered diving into the different tiers and feature availability. Don’t forget that we’re basically dealing with the equivalent of an arms race, so new features and add-ons and capabilities are just constantly being added on. From the time of recording to the time of publication, which will be less than a week by the way, who knows what might change? 

But for now, ChatGPT seems to be on the leaderboard as the all-around queen of the AI tools. It does pretty well in content generation, coding, fact-checking, and reasoning. Want more polished prose? Try Claude. Going for thorough research? Try Perplexity. Looking for a more left-brained approach to problem-solving? Maybe DeepSeek is what you seek. But maybe you just want something that easily and reliably integrates with your existing Google ecosystem? Well, you know which one is best for that.

All that to say, listeners, there really isn’t a one-size-fits-all solution when it comes to LLMs for marketers. Each tool we’ve talked about today brings its own unique strengths to the table. 

In our opinion, it’s probably a good idea to combine these tools whenever you can. But just remember that your trusty AI friendI is here to assist, not replace, human creativity and critical thinking. 

I think we’re wrapping it up now. Thanks for tuning in to this brief episode of SEO Unfiltered. Don’t forget to subscribe and leave a review. You can also find us on your go-to podcast app or visit our website at geekytech.co.uk.

Until next time, keep your content fresh, and your SEO strategies fresherer. Thanks for listening guys.

Peace!