If you’re a marketer who’s been meaning to actually learn the difference between ChatGPT, Gemini, and Claude—but keeps putting it off because it feels like homework—this one’s for you.
I’m breaking down the major AI platforms side by side: what each one’s good at, what the different models actually mean, and which one might fit how you work. No hype, no jargon, just a practical look at what’s out there so you can pick your tool and start using it.
Whether you’re trying to speed up daily tasks or think more strategically about AI in your workflow, this is your no-BS starting point.
00:00 AI LLM models and modules
02:58 ChatGPT: Deep Thinker
03:32 ChatGPT: Quick Responder
04:09 ChatGPT: Multimodal
04:39 ChatGPT: Code Optimize
05:27 ChatGPT: Tool Enabled
05:59 Google Gemini Pro
08:16 Gemini Creative AI Modules
08:22 Gemini Nano Banana for Images
09:02 Nano Banana Pro
09:44 Gemini VEO Series for Video and Audio
11:01 Gemini Live for Mobile
11:30 Gemini Personal Intelligence as a Personal Assistant
12:08 Claude AI Pro
13:52 Claude Haiku 4.5 (the Speed Demon)
15:09 Claude AI Sonnet 4.5 “Goldilocks”
15:23 Which Claude model to use for marketing work?
16:15 Claude Opus 4.5 “Deep Thinker”
AI: 4 AI LLM models and modules
[00:00:00] Rachel Moore: You’re going to get asked, not if you use AI, but how you use AI. The assumption is going to be you’re using it. Okay. My recommendation to you is that you do take, take a day. You might not even need the whole day, but start small. If you are using AI already and you’re creating prompts, and you’re prompting AI to help you with something, open one or two additional tabs.
[00:00:30] And in each of those tabs, open an alternate AI platform. So if you’re used to using ChatGPT, great. Open a tab, open Gemini, open another tab, Claude, Perplexity, something else out there like that. Enter the same prompt and see what you get. Compare and contrast.
[00:00:55] Rachel Moore: In this video, we’re going to go through the different chat [00:01:00] modules or modes that are in three of the kind of mainstream AI chat platforms. We’re gonna look at ChatGPT free. We’re gonna look at Google Gemini Pro, and we’re going to look at Claude AI Pro. So we’re gonna kick things off with ChatGPT, but.
[00:01:19] Real quick, I just wanted to let you all know, so what you’re about to see as I share my screen is that I entered the exact same prompt into all three in individually, and I’m gonna show you what it came up with, which will help guide us. What better source to ask than the actual chat platform that we’re using?
[00:01:40] So first up, we’re gonna look at ChatGPT. So I’m gonna show you the prompt that I used. I said I’m about to record a video describing the different LLM modules or modules available in different AI chat platforms. Wanna help me with yours, ChatGPT? And of course, it answers in its own little way. And by the way, when we were doing this, I said I’d like it to be [00:02:00] quirky and fun. That tends to be how I like to communicate with my chats. You can think of it like, so it’s saying ChatGPT isn’t one brain. It’s more like a cabinet of specialists. Different models are tuned for different jobs. In last week’s AI imposter syndrome video, I refer to this a little bit, that one way you can think about AI is like you are a boss or an editor in chief, and you might have different writers on your team, different workers who have different beats, and different tones and ways they write, different ways that they approach things and how they deliver those things.
[00:02:37] So think of it that way. You can actually, in one platform, have that slew of different employees, if you will, or direct reports, who you can say, okay, I’m gonna ask you to figure this out. Okay, now I’m gonna take it over here. I’m gonna ask you the same thing, and you’re gonna get a slightly different result or maybe a drastically different result based on who you’re asking.
[00:02:58] ChatGPT: Deep Thinker
[00:02:58] Rachel Moore: So first of all, we’ve got [00:03:00] in ChatGPT, calls it the flagship reasoning model, the Deep Thinker. That’s the one you’re usually talking to by default. It’s good at complex reasoning and problem solving, strategy, planning, writing that needs some structure, coherence, teaching, and explaining concepts.
[00:03:15] It can be a bit slower, uses more computing, and might be overkill for simple questions. This is the model you want when you care about thinking, not just answers. So this isn’t just say, Hey, where can I find a water heater? Probably not that. You instead wanna be like, can you tell me how a water heater works?
[00:03:32] ChatGPT: Quick Responder
[00:03:32] Rachel Moore: Then you’ve got the fast, lightweight model: the Quick Responder, good at quick Q&A, rewrites. As a writer in marketing, and again, this will deal directly with marketing. If I’m like, ah, I need to write an email. Maybe I’ve got a blog copy or a case study copy, and I need to turn it into a marketing email.
[00:03:51] Okay. Go rewrite the context of the case study and rewrite it into a marketing email. Brainstorming, chatty back and forth. Bouncing ideas off. So this might be best [00:04:00] for a marketer to use. Less deep reasoning. It might miss some nuance. If the flagship model’s a chess grand master, this one plays blitz.
[00:04:09] ChatGPT: Multimodal
[00:04:09] Rachel Moore: Then you’ve got the Multimodal Model, the can see and hear one, that one handles images, screenshots, diagrams, and sometimes audio. So if you’re not just like inputting text as a prompt, it can analyze images or documents, explain charts, photos, UI screenshots, combine text and visuals in one conversation
[00:04:29] Just know visual interpretation can still be imperfect, not always needed unless you’re working with media. That’s where it stops being text only and starts interacting with the real world.
[00:04:39] ChatGPT: Code Optimize
[00:04:39] Rachel Moore: Then we’ve got Code Optimize model. Now this gets into, and I work with a client where we have, it’s a B2B SaaS company, and I interact directly with software developers and engineers.
[00:04:48] This would be what they would be using to help with coding and stuff. Vibe coding, if you’ve heard of that. Writing and debugging, code explaining errors, refactor. That again will make sense to [00:05:00] software developers working across multiple languages, including programming languages. It’s not as expressive for creative writing, less conversational.
[00:05:07] So I would not necessarily use this for marketing, but unless I’m trying to market for those engineers and my audiences engineers, maybe I’d wanna incorporate this one a little bit. Same ai, different training priorities, fewer poems, more pull requests. If you don’t know what a pull request is, a software developer near you knows what it is. It’s a GitHub term. alright.
[00:05:27] ChatGPT: Tool Enabled
[00:05:27] Rachel Moore: Then there’s Tool Enabled mode, the AI with hands. it’s not necessarily says a separate brain, but important concept. It can do web browsing for updated info, file uploads and analysis, data processing, image generation, external tool use. The model reasons, but tools let it act.
[00:05:43] This might be one where you’re working more with APIs that connect with your ChatGPT and connect tools that you use.
[00:05:49] Okay, so that’s a super quick rundown of things and capabilities, different models and modules you can use in free ChatGPT.
[00:05:59] Google Gemini Pro
[00:05:59] Rachel Moore: [00:06:00] All right, now we’ve switched over to Google Gemini, and I’m using Pro, and I do that because I have Google Workspace, that’s the operating kind of tech stack system I use for officey stuff: email, calendar, whatever, versus me using say Microsoft Office. As a result, because I pay for Google Workspace, I also get Google Gemini Pro
[00:06:22] Just know Gemini free will probably have some limitations to it, but as I promised, I entered the same prompt that I did in ChatGPT: I’m about to record a video describing the different LLM models or modules available in different AI chat platforms. Wanna help me with yours, Gemini?
[00:06:40] So it’s giving me the core language models. The current generation is the Gemini 3 family. The main choice for users is between raw speed and deep reasoning.
[00:06:50] This should sound a little similar to what we just went over with ChatGPT, but this will also give you some options to say, what do I need based on what I need to get done?
[00:06:59] And this will give you [00:07:00] Gemini options. So Gemini 3 Pro is the “thinking” model. It is designed for complex reasoning, heavy coding, and PhD-level math. It’s the powerhouse. Okay? So again, software developers would wanna use this. If you’re trying to come up with a very complex maybe marketing strategy and incorporate a bunch of different nuance into it, this might be your go-to, and of course, math. I’m not touching that, but if you are trying to do maybe some metrics analysis, maybe here too.
[00:07:29] Gemini 3 Flash: the “fast” model. This is the new default for the Gemini app. It provides pro-class reasoning, but at lightning speeds, making it ideal for everyday tasks and real-time interactions.
[00:07:40] All right, and this would be the one marketers probably where if you’re dealing with copy, you’d use Gemini 3 Flash. You’re like, Hey, help me write a series of social media captions, marketing emails, develop a case study with me. I’ve got a transcript of an interview. I did a video interview I did with a client or a customer, take that and let’s turn that into a case [00:08:00] study, and I can give you a template for it, and here’s how you’re gonna structure it, kind of thing.
[00:08:04] Now, phasing out Gemini 2.5 Pro Flash. It remains available for specific workflows like the Canvas Interactive workspace, but it’s being phased out in a favor of the 3.0 series we already talked about.
[00:08:16] Gemini Creative AI Modules
[00:08:16] Rachel Moore: Then we get into creative modules. I think it just came out, I wanna say November, December, 2025, the
[00:08:22] Gemini Nano Banana for Images
[00:08:22] Rachel Moore: Nano Banana series in image generation and editing. So this is a separate engine for that multimodal system that doesn’t just make images but collaborates on them.
[00:08:32] And I’ve had hit or miss with this. I tried it with great success to create an image for me, but then I, using the same image, it was my headshot, I was trying to have it do something else, and it was like, anyway, you want. So that might have been user error with prompting. I’m just saying Nano Banana Standard is powered by flash architecture.
[00:08:50] I meant the Gemini 3 Flash up here. It’s best for fast iterative edits, style transfers, and doodling your changes directly onto an image, so it can help you gen [00:09:00] generate images simpler ones.
[00:09:02] Nano Banana Pro
[00:09:02] Rachel Moore: Nano Banana Pro is powered by Gemini 3 Pro. We talked about that up, up top here. It handles complex scene logic, 2K resolution, and high-fidelity text rendering for things like logos and posters.
[00:09:13] Key features are character consistency, keeping the same person in different scenes, and the ability to combine multiple lo uploaded photos into one new composition. So do play around with that. Use AI for your creation and marketing, but I would always let people know I’m using it. Maybe not try to hide the fact I’m using it.
[00:09:32] It’s getting better at where it’s harder to tell. I always look for that sheen on people where it’s like a human, and I’m like, this looks really AI ‘cause they almost look like a Barbie doll. It’s got that like plastic sheen that’s starting to go away more and more.
[00:09:44] Gemini VEO Series for Video and Audio
[00:09:44] Rachel Moore: So video and audio.
[00:09:46] There’s the Veo series for video generation. Veo 3.1 is now directly integrated in Gemini app and YouTube Shorts. Veo 3.1 can generate high-fidelity video up to 4K with native audio. If you prompt for a [00:10:00] cat playing a piano, it doesn’t just show the cat. It generates the synchronized piano music, creepy.
[00:10:05] Ingredients to video. You can upload ingredient images like a character or a background to ensure your video stay stays consistent. It is mobile optimized for 9:16 vertical video. It integrates into YouTube Shorts. YouTube is gonna ask you if AI was used to generate a video. It has a checkbox in there.
[00:10:22] Please be honest. Just be upfront and say you did it. And maybe if you’re just like feeling squeamish about that, say, look, this is what I did. I created this with AI. Isn’t this neat? Use the wow factor. But you should be upfront about it. There are obviously people out there who are gonna use this for nefarious purposes.
[00:10:39] Don’t be one of them. I will leave it at that. Don’t be fake. Obviously, if you’re using this for marketing based this on reality and truth, you can absolutely use it to generate videos. Make sure you market accordingly. Self-proclaim, self-disclaimer that you have used AI and don’t base it on erroneous fake [00:11:00] shit.
[00:11:00] Okay?
[00:11:01] Gemini Live for Mobile
[00:11:01] Rachel Moore: Then there’s special mode Gemini Live conversational mode for mobile. So you can use this on your mobile app, real-time, interruptable voice chats, and can see through your camera to help with real-world tasks. So you could have your phone and be like, okay, I have an ingredient, or I’m seeing this vegetable maybe at the grocery store.
[00:11:18] Show it to Gemini Live on your phone, and it could probably tell you, oh, that’s a rutabaga, I dunno what a rutabaga looks like. Or maybe you’re like shopping and you’re like, Ooh, what kind of car is that? I’m not sure. And it can show you.
[00:11:30] Gemini Personal Intelligence as a Personal Assistant
[00:11:30] Rachel Moore: Personal Intelligence is a beta feature for Pro Ultra subscribers that allows Gemini to securely connect to your Gmail photos and drive to act as a personal assistant. I actually have this. I have not really scoped this out, frankly, it’s because I use Claude Pro, Claude AI Pro.
[00:11:46] I prefer it. This is all gonna be based on what kind of results you get that deliver the best for you. Please do play around. I highly recommend trying the same prompt with the same [00:12:00] material and input and into two or three different chat platforms like we’re doing today. See what gets you the best result based on what you need.
[00:12:08] Claude AI Pro
[00:12:08] Rachel Moore: Finally, we’re going to take the same look over at Claude AI. I was a hard sell on that one because it was being used primarily by, again, I work with software engineers and developers.
[00:12:20] They were using Claude Code, but I’m like, let me dabble. I like it a lot, and I do pay for it. Yes, I have Gemini through Google Workplace. I prefer to use Claude. You may have a different favorite, completely fine, based on what you need, but here’s what I did.
[00:12:36] So, as you can see from my screen, I’m using Pro, I’m doing it within a project where it is helping me plan and ideate the how-to videos I do for marketers like this one.
[00:12:49] I love that it is like wait a second. Let me look contextually at that before I answer your question specifically. It gave me some video ideas for what I can talk about, [00:13:00] which y’all, I’m totally gonna be like, okay, good. Let me grab some of these for future videos. So now you’re getting a little bit of a teaser sneak peek, what might be coming on some of these upcoming AI videos, building content pillars with Claude.
[00:13:11] Woo. Let’s talk about those. Campaign brief can place nice, but now we’re talking about LLMs, and it got a little bit into it. I said no, I need to get to the level of specifics that I got from the other two from ChatGPT and Gemini. So notice that I gave it additional prompt. Can you specifically define for me the differences in your models, like how Haiku 4.5 is different from Opus 4.5, et cetera?
[00:13:35] Alright, so now found me the specs. This is what I’m gonna go over with you. Again, getting through Claude, knowing that it’s got this capability.
[00:13:42] All right. The three Claude 4.5 models, and again, I am recording this in January 2026, so if these update, we can do a new video.
[00:13:52] Claude Haiku 4.5 (the Speed Demon)
[00:13:52] Rachel Moore: Claude Haiku 4.5, which is what I used for this particular query, the Speed Demon. Lightning fast, almost instant [00:14:00] responses cost $1 for every five per $5 or one $5 per million tokens. Input output. Okay, so $1 per million tokens input $5 per million tokens output. Context, one to 200,000 tokens.
[00:14:13] You know what my next question’s gonna be for these? What are you talking about? That should be my next video. How do these things charge? Like by tokens or the API hits or whatever?
[00:14:23] Best for quick tasks, high volume work, rapid iteration, customer service, chatbots and social media caption generation. So remember, it’s bringing this into context.
[00:14:34] It’s taking this from context that I’ve said how to videos for marketers. It knows that’s what I’m asking about. The answer would be different if I came into Claude AI cold turkey with no context, saying, tell me about your models. I probably get more similar to what I got for ChatGPT. Remember, I was actually saying, Hey, based on what I’m reading here, this makes sense.
[00:14:55] It makes sense for marketers. It’s already giving me that context here. So just know that it’s [00:15:00] coming into that, real talk. It’s your choice when you need snappy answers and iteration. Speed matters more than perfection. Perfect for brainstorming multiple angles fast.
[00:15:09] Claude AI Sonnet 4.5 “Goldilocks”
[00:15:09] Rachel Moore: Then we get into Claude Sonnet 4.5. Goldilocks. Fast, responds within seconds. Little bit more expensive per million tokens. Context window, 200,000. Daily work, complex reasoning, coding, content strategy, competitor analysis, multiple-step projects.
[00:15:23] Which Claude model to use for marketing work?
[00:15:23] Rachel Moore: So this is more like I said, Hey, I’ve got a case study, transcript from a video interview I did with a customer using that, can you just draft me up a series of marketing emails? Here would be Claude Haiku 4.5. Hey, I wanna come up with a content strategy of different kinds of case studies we could do, but also how that would incorporate into more integrated marketing campaigns we’re gonna do through 2026. And maybe for H1, I wanna focus on, financial industry. For H2, I wanna focus on legal. Help me devise [00:16:00] that. You’re gonna use Claude Sonnet 4.5.
[00:16:02] Okay. This is Anthropics official recommendation for most people. It’s where you’ll spend most of your time. Bonus: it excels at creative content tasks like presentations and animation matching or exceeding Opus 4.1.
[00:16:15] Claude Opus 4.5 “Deep Thinker”
[00:16:15] Rachel Moore: Okay. And then we get into Claude Opus 4.5. Notice it says 4.1. That’s prior to 4.5. This is the deep thinker, though moderate speed. It takes longer, but applies deeper analysis. And it’s more expensive for a million tokens. Context window, unique feature only model with an effort parameter where you can control thoroughness.
[00:16:35] Set to high for critical work, medium for prototyping, low for quick exploration. I’ll be honest. I tend to have it set to Opus 4.5. For complex work, Sonnet. Again, Sonnet would be great for those, like, hey, email copy based on a case study or something like that. But I have done 4.5 for when I’m talking about doing these how-to videos because [00:17:00] I tell it like what my end game is.
[00:17:01] I’m like, I wanna really explore the efficacy of Substack for marketers. I wanna see if I can monetize my Substack. I wanna monetize my YouTube channel. I wanna be really going hard, which is, this is all true. In 2026, I am testing the waters of Substack for marketing, of YouTube for monetization and marketing, and I’m coupling all that.
[00:17:24] The third leg of my stool is LinkedIn, my personal profile on LinkedIn. I’m trying to see how those all work together and if those all do work together for me, for the balance that I’m doing, I’m using Opus 4.5 to sort that out. Complex reasoning, high-stakes decisions, architectural problems, novel problem-solving, so that’s what it’s providing there.
[00:17:43] I’ve said this throughout each of these kind of demonstrations, these screen shares, where we’re digging into those different models and things like that. I spoke to somebody who went to school for linguistics, and therefore they have really great insight into language learning [00:18:00] models and how linguistics completely figures into that and feeds and helps these things build their knowledge.
[00:18:08] One thing, I’m gonna leave this with you and all of this, the crux of this is for you marketer, marketing professional. And if you’re not a marketing professional watching this, please do take this, it’s probably gonna apply to you too.
[00:18:21] Your future career,
[00:18:23] Rachel Moore: You’re going to get asked, not if you use AI, but how you use AI. The assumption is going to be you’re using it. Okay. My recommendation to you is that you do take, take a day. You might not even need the whole day, but start small. If you are using AI already and you’re creating prompts, and you’re prompting AI to help you with something, open one or two additional tabs.
[00:18:54] And in each of those tabs, open an alternate AI platform. So if you’re used to [00:19:00] using ChatGPT, great. Open a tab, open Gemini, open another tab, Claude, Perplexity, something else out there like that. Enter the same prompt and see what you get. Compare and contrast.
[00:19:15] Do this. Start small, start at least once a week, and make sure that the prompt, as you do each experiment is different.
[00:19:22] You’re asking for something different. So what I’m saying is maybe this week you are trying to develop some social captions for a blog or a series of blogs, and that is your prompt on two or three tabs. Next week, maybe you’re drafting a marketing email for an event. Again, that’s your prompt on each of the three, and then the next week, go more complex. What if I wanted to come up with a six-month content strategy that’s incorporating our social media channels of LinkedIn company page, Facebook page, and YouTube channel, an email drip campaign, four case [00:20:00] studies, and a webinar.
[00:20:02] Pop that into those two or three AI chat tabs and see what you come up with. And again, ask it when you’re entering the prompt. Which model should I use for this? Which model would you recommend? Remember, with prompts, we’ve talked about context on prompts in some of these past videos.
[00:20:20] You need to tell it who it’s supposed to be for you. Are you a super strong coworker that knows marketing and is familiar with this brand that I work for? You can give it a lot of context and cues, so just make sure you’re doing that. But you will get asked at future jobs, how you use AI and won’t it be great when you can say, well, I prefer to use Google Gemini for this exercise or this series of tasks, tasks, or tactical activities for, for here, but I then I switch over to this model. Or you know, ChatGPT model for when I’m developing a content strategy, I’ve found [00:21:00] that I get that better. Any kind of response like that will poise you better than other candidates because at least they’re like, you’ve tried this out, you are playing around with this in seeing what works best.
[00:21:15] Great. It’s as crucial right now as it is to say, do you freaking know how to use the tools of your trade, like email platforms and social media apps and how to write blogs. It’s right up there. Okay. It’s as crucial as that stuff, as the rest of the tech stack. So is this so, test and try.
[00:21:38] I would love to hear what your favorite, not only chat platforms are, but models, modules, which ones you’re like, oh my God, I love Nano Banana. Tell me about it. you could help guide somebody else who’s got the same, like, well, I need to sort this out. You liked that one.
[00:21:53] I’ll try it next time. Okay. Let’s crowdsource this a bit. Thank you so much for watching, [00:22:00], and good luck with your next series of AI prompts, testing, and playing out the modules.
[00:22:05] Rachel Moore: You’re still here. It’s over, but you should hit like on this video before you head out and subscribe if you wanna get “moore” videos that’ll help you realize how much “moore” there is to marketing.
[00:22:16] See what I did there?
But wait! (Yep, there’s more for Substack subscribers.)
Exclusive content is available with Rachel’s livestream videos: Boss Level Mondays and Hot Mic Fridays. Start your week off with irreverence and real talk as you face down the work week of a marketer with Rachel at your side. Then, on Fridays, head to Hot Mic Fridays for unfiltered takes from the week, as well as podcasting and video tips you can always use. Subscribe to Rachel’s Substack to access and join Rachel live!