Minimax 2.1 VS Claude 4.5 VS Gemini 3: Who Wins?

We are going to be testing Miniax 2. 1 versus Google Gemini 3 versus Claude Opus to see which is the best AI coding

tool. Now, if you've never used Miniax 2. 1, you can actually get access directly on miniax. io and this is a new powerful AI agent that's great for coding, right? You can actually access it via API and you can access it inside the agent here. Now, when you use it inside the agent, you can get access to the videos and the content creation. So, you can create dashboards. You can build cool real tools like you see right here. This actually created a wall of testimonials scraping my website as you can see. But also going to compare this against Gemini 3 Pro and Claude. Now if

you actually have a look at Miniax on the benchmarks here, this Chinese AI tool is pretty comparable on the benchmarks, right? It's very interesting to see that it's actually competing with some of the biggest tools in the world, but it's open source and you can run it locally, right? So you can see here for example Miniax 2. 1 and you see how it performs versus Gemini 3 Pro and Gupty 5. 2 Thinking and Claude Opus. Now what's interesting here is if we have a look here we've got MultiSswe benchmarks right and Minia 2. 1 scores 49. If you have a look at the benchmarks here Gemini 3 Pro only scores 42. 7. Claude Opus is pretty much on par with Miniax at 50 and G5. 2 doesn't score in that. Okay. And you can see here also you got SWE bench multilingual 72. 5 versus 65 and 72. So it's pretty interesting to see that these benchmarks are very close, but I never really pay attention to the benchmarks to be honest with you.

What I really care about is how do these models perform in reality. All right, so we're going to be comparing these models side by side, coding out with them, running a coding challenge, and seeing who performs the best. All right, so

we're going to run this on open router to use the API. If I use the agent directly inside the chat, it's going to be super slow, right? That's why I'm not using the agent inside the agent section. And I can access the API directly here, but I think the easiest way to use it is inside open router. Sporter says this should be a good showdown. I'm excited for this one, too. And we're going to be testing this out.

All right, so we're comparing Miniax versus Claude versus Gemini and seeing how they perform. Now, if we go to Open Router, we're going to grab some of the prompts from today. But basically, here's what you're going to learn today. All right. So, you're going to see how to use all three of these. By the way, I promised to show you how to use them for free. Right now, if you want to use them for free, what you want to do is download anti-gravity and you can use Cordopus Gemini 3 for free. If you want to use Miniax 2. 1 for free, then what you want to do is check out Olama and then run Miniax on cloud. So, if you get here and then you type in Miniax, right, you can see Miniax 2. 1 is available on cloud when you run. And that's how you can run it for free. So you can check out all these models for free. We're going to look at which one is best for building apps, writing content, which one is the best for writing coding agents, which one saves you the most time, the exact prompts I use for each test, and how to pick the right AI for your business. All right, we got some good banter inside the chat here. Scully says Goldie, just in time for the showdown. Border says Scully and Scully says BP. So we got a good banter going on here. Good to see the community going. All right, even if you've never heard of Miniax before, even if you think AI coding is too hard, even if you're confused about which AI to use, by the end of this stream, you'll know exactly which AI to use for what, because we're going to test them out side by side. This is called the Goldie Battle framework that we're using. This my simple system for picking the right AI for the job. I call it the battle framework. All right, so build test is what B stands for. So, can it build a working app? Agent test is can it use tools and browse the web? Time test is how fast does it respond? Truth test is how accurate is it? language test is how good is it at writing effort test how much back and forth does it need all right we'll score each AI on these things separately today we're going to start off with these tools all right and so we're going to go on to the AI now and we'll test these out side by

side all right so we're going to go inside MAX and we'll say tell me one thing about yourself that no other AI can do all right and we'll do the same inside Gemini and the same inside called Opus 4. 5 and actually I should be using pro here so I'm actually going to switch now there we go all right let's stop that run again. So, tell me one unique thing about yourself that no other AI could do. Let's see what we can. Scy says it's like watching the playoffs. I'm glad I'm on the same level of entertainment, mate. Thank you. So, you can see here, for example, it says, I'm trained by a Chinese AI company. I have strong blah blah, one unique characteristic, my specialist tuning gave me some patterns and approach blah blah. All right. And it says, "The most honest answer is that it comes from a uniqueness combination of training data, particular response style, cutoff, and my developers particular approach to AI safety. " Right? So, I quite like the way that it's displayed. The way that it writes is straight to the point, no fluff, etc. And Gemini is to be honest, mate, claiming that I can do something no other AI can do is pretty difficult because the field advances so rapidly. A feature is today that might be industry standard today. All right. And then it actually says, "My massive working memory is a big unique point. " And I would agree with that, right? So for example, Gemini has a context window of 1 to 2 million tokens and I think it's 1 million. Whereas for example other AIS like for example Claude only has a context window of something like 200k tokens, right? And so that gets annoying. Now if we actually have a look at Miniax, you see Miniax inside let's check out an open router max 2. 1. So Miniaax's token context window is 205,000 tokens. Claude is 200,000. And so Gemini genuinely has given the best answer for that so far. Let's have a look here. Claude is this is a terrible answer. It says one thing that I can that's genuinely distinctive. I can sit with uncertainty and nuance without rushing to resolve it. It's just what are we talking about here? I would say that Claude probably gave the worst answer. I would say that Gemini gave the best answer and Miniaax's was mediocre. Right. And Sporter says Minia tough but he's on a llama. Not sure it can hang with Gemini. Yeah. So I'm going to use it on open router which means I'm using the API directly. So, we should get the best answers from that, the best outputs. Next up, let's do a simple

writing test. So, we're going to say, "Write a three sentence description of an SEO agency that helps businesses get more customers through Google. " All right, plug that in. Here we go. Scully says, "I do get annoyed with calling the context window. " Yeah, one of the good things they actually solved it recently is like they can compress the memory inside the chat so it never really runs out of context window if you're on the premium plan, but yeah, for sure. Like before when it would just break and stop working, that was a nightmare. So, it says we help. Okay, so that's one example. Let's see. Gemini, blah blah. Let's have a look. Mini Max. So, we help businesses attract more customers through proven search engine optimization. These are all pretty boring responses. I don't really like that one so much. This one just feels like AI generated fluff. And then, Claude, we help business shop first when customers search Google for what they sell. All right, that's probably the best description. I would say that Claude has won that just by a little tiny margin. I wouldn't say any of them were particularly impressive there. Scully says, I have to go for Claude on writing. I would agree with you. I think that Claude won it again. So, just to recap here, we'll put this on a little table here. So, Gemini first for unique thinking, right? Claude came first for writing. Let's see what else we got here. So, if we go back to the prompts, I'm going to skip through some of these days just to give something more interesting. So, I'm going to go to something a little bit more interesting. I'll say create a Pomodoro timer app with 25m minute work sessions, 5 minute breaks, counter for complete sessions, and sound alerts. All right. So, we'll copy that in. Go to I could use the agent for that but it's just going to take too long. Do you know what? We'll do it. If it doesn't perform in time, then we're going to switch to interactive app over here. We'll do this inside a new chat on Gemini using canvas and we'll do a new chat inside Claude. There we go. All right. So, these are all coding out. We've got the agent. We have open router. We have Google Gemini. We have Claude as well. And we'll see which one performs the best. So, Gemini has created something pretty nice right there. That's working nicely. We can reset that. We can skip it. It counts how many sessions we've completed. It's super nice. Plus, we can easily share this as well. That is good. That is a good one, my friends. If we go to Claude, let's have a look here. Claude is still coding out and Miniax is already finished. So, let's have a look here what we got. We go in the preview, see if that's worked. I don't think that's worked at all, is it? That is no bueno. Now, having said that, we do have the AI agent building out on the left hand side. So, we'll wait for Miniaax agent. That should be done pretty soon, I think. If you go back to Gemini, I think Gemini has created something really nice right there. Let's have a look at this is Claude. So, we can skip between them. We can reset the timer. It works perfectly. I would say the UI is still nicer on Gemini. This looks a lot nicer than Claude's output. Both of them are functional, but one of the interesting things here is like it actually counts how many times this has been completed. So, how many Pomodoro sessions you did. And I would say just by the tiny little bit, Gemini 3 Pro has won it again there. So, let's put this in. So, we're going to put Gemini, one for the Pomoduro app. And so, we've completed three tests so far. Gemini has won two of them, and Miniax doesn't want any of them. Claude came first for one of them. All right. Now, we've actually got the completed output from Miniax agent over here. So, this is the agent. But look at that. The 25 doesn't even fit in the circle. We can switch between break and work. I do UI, but why is he got why is that not inside the circle? So even like the fundamentals of coding are being missed there. So again, I would keep the rankings as the same because Claude's UI was much nicer. Let's see what we got. Sporter says, "Would AI Studio perform better or is 3. 0 the same? " I tend to find that AI Studio it creates more complex output sometimes. It's more designed for building apps, but 3. 0 is much faster and easier to deploy stuff, right? And it doesn't really require any technical skills. So far, you can see how we've done. Let's go on to the next prompt. We're now going to say, let's see what we can do here. Create a simple blog website with homepage. I'm actually going to go with something more interesting. So, I'm going to say build a hyperbond game, right? Make the ball trail neon spark. Speed it up with each bounce. Explode it into particles on every point score. All right. This is going to be a bit more interesting. So, we're going to go into miniax 2. And bear in mind like I'm rinsing the credits here. We're smashing through the credits. We'll run that. Then, we'll go over to open router. Start a new chat. We'll do this on Gemini. And we'll do this inside Claude. All right, I'm back in the game, peeps. Back in the game. So, if you have a look here, we've got the hyper pong game, isn't it? That's looking super nice, right? Oh, there we go. Incredible. What a time to be. Ah, yeah, we scored a point there. Don't worry about that. This guy's hard to beat. Anyway, that's very addicting game. That's pretty fun. Have a look at Minia Max. It's finished. All right, so we're going to open this up and we check out what we got here. We'll go to the preview. Why is it not allowing me to preview? Let's open this up in Why is it only created that HTML? That doesn't look right to me. Let's have a look at live weave. And that is no bueno, is it? We'll give it another chance. So, it's building out on the agent as well. I do think the agent is way better from what I've seen for the outputs, right? Way better so far. So, I think honestly, if you want to get the best out of Miniax, you're probably going to use the agent, not the API, or you're going to use the API in something like Visual Studio Code, which is more like trained to guide it because inside open routed, the output. So, really bad. We're waiting for that to load. Waiting for Claude. Oh, Claude has come back to us. Now, I do think that Gemini is going to be hard to beat here. So, let's have a look. Oh, look at that. But we got an error inside Claude. It broke straight away, my friends. Pawn game. Gemini has won again. And so you can see Gemini, right? Like it's absolutely crashing. We'll wait for mini match to come back, but so far it's coming in last cuz it didn't even create something that works. Let's keep going through now. So leave that running in the background. So now we're

going to do a creative writing showdown, right? And we're going to say write a story about a solarreneur who discovers AI transforms their business. Make it inspiring but realistic. And hopefully Miniaax gives it itself a chance to redeem itself. But let's see. So we're just going to go into normal writing mode here. Yeah, Gemini. It's just a good game. Like I can't stop playing it. We're going to X off of that. And then we'll do the same inside Gemini. Clawed obviously broke. Okay, great. But we'll try this out. Minia says it's like they're letting you fly the plane for free but making you pay for the fuel. That's a good analogy. I like that one. All right, so this is the story test. Let's see what we got here. I like the way it formats it. Like it does look nice on the formatting. So it says Mayia Chen had built her graphic design business the hard way nights hunched over the laptop weekend sacrifice client revisions in an inbox that never stopped demanding attention and then it talks about the accident discovery quite let's have a look at the end oh it's a long one that is blasting out a lot of words isn't it that is a long story let's have a look at Gemini see this is way more interesting just even as a hook right so for example if you have a look at the hook right the first line where it says hen had a graphic design business the hard way nice blah blah Right. Straight away it's got those M dashes which just make it harder for my brain to read. But also number two, it's like quite a boring introduction to a story. And a story should grab people's attention straight away. Right? The story is only as good as the attention it captures and maintains. If we have a look at Gemini, it said Elena stared at the cursor blinking on the screen. It was 1:47 a. m. Right? Her business blah blah was technically a success, but she also had zero blah blah. Right? And so that's quite a nice story. Quite a nice interesting hook. If we have a look at this one, do you know what's interesting is like this one has said Mayor Chen and Claude said Marcus Chen, right? They've used similar characters even though the prompt doesn't mention anything about someone called Chen. That is strange, isn't it? And then if we have a look here, Marcus, who've been running this one man marketing consultancy for 6 years, no blah blah trapped blah blah. I think honestly the hook is way better inside Gem. So I think for creative writing, Gemini has probably won that one. Let's have a look at the last line. She walked out the door not to a meeting but to a pottery class. She was an artist again. Let's have a look at the last line of Gemini. The AI hadn't replaced him. It removed the ceiling. It still worked out. The tools didn't change that. Okay, that's a little bit boring. And then I like the ending with the pottery class on Gemini. And then this feels more like a corporate block, right? Again, I would say that Gemini won for the creative writing as well.

And so if I had to pick any of these models, of course, I'm going to go with Gemini, right? Gemini one for unique thinking, the Pomodoro app, the Pong Game, and creative writing. And I'm not going to switch from MIA anytime soon, but at least you know how to test this stuff and to check it out and how to create things with this, right? And if you want all the prompts from today, you can get it inside the AI success lab completely for free. All right, that's interesting. Scully says, "Minia Max use Men. I've seen a name many times in the copy. Marcus Chen is the other one. Cord will use in storytelling as well. " Interesting. All right, so for picking an AI tool, let's talk about the old way and the new way. Old ways like use whatever AI is training on Twitter. You stick with one AI for everything. You pay for subscriptions just in case. You ask friends which AI is the best. You read reviews and never actually test. Right? That was the old way of testing this stuff. Now what I'd recommend is as you've seen today, test everything on your actual tasks. Use different AIs for different jobs. Right? So for example, if you go over here, I probably use Gemini for unique thinking, but for writing, I'm going to use Claude. Right? For coding apps, Gemini. For coding games, Right? So different AIs are better at different things. And then you can start with the free versions first. So for example, Miniax, Opus, and Gemini all have free options. You can score each AI on what actually matters to you. And then you pick based on results, not high rank. So that's the way that I

would look at it. Yeah. And if you want to get all the notes from today, feel free to get that inside the AI success lab. Completely free community link in the comments description. Connect you with 40,900 members. As you can see, you can post inside the community, get help, get support, that sort of thing. You can post inside the classroom here. And you can get access to all the video notes as you can see inside the classroom right here. Now if you want to get personalized coaching, if you want to connect with me, if you want to get a supportive community that raises you up, then join the AR profit boardroom link in the comments description. Inside here, you can get access to this amazing community where you can post questions, get help, get support, and also inside here we have a win section, right? So for example, Justin is celebrating the growth of the members inside the group. Thank you very much, Justin. Porsche was talking about her chat cheap surprises with the yearly update. You can see Scott is sharing his 100 day journey. You've got, for example, we actually post appreciation posts for the best members and the most active members and that sort of thing. You can also see Karen, for example, made a few client websites, built a custom GP and created one out of 50 tools that she's building out right now. So, loads of cool inside there. And then also inside the community here, you can see we have an accountability group, right? So, we have this accountability group and that turns intention into action. So, one of the things that a lot of people struggle with when it comes to AI is they really want to use this stuff, but they don't know how. And so, inside the community, what we do is we have a daily accountability group thanks to people like Darren. And our members just create this group where they share their goals, they share what they're working on, learning, and then we all help each other, right? So, for example, Justin explained some of the goals that he's working on. And these can be personal goals and that sort of thing, but the main thing is we're just learning, we're growing, and we're sharing what we're working on together as you can see, right? So, Porsche, Carlo, etc. We're all joining in and helping each other. And also inside the calendar here, we do four weekly coaching calls. So you jump on the coaching calls, get help, get support whenever you need it. And then inside the classroom here too, you can see that we have a six week AI automation masterass, which shows you how to go from a complete beginner to becoming a master inside AI automation, plus how to build your first AI agent in less than 5 minutes. And on top of that, we have all of these different trainings, playbooks, etc. everything that I use to automate my business with AI. So, if you want to learn about AI automation, you want to use what's working for me right now, you get that inside the classroom along with our best courses on A10, how to get clients, our coaching call recordings if you miss them, you can also get our best AI automation and also YouTube, right? So, we have a sixe road map on how to use YouTube and how to create as well. All right, Wesley says now here working on something listing. Good to see you here. Wesley, good to see you. Isaiah says, "Are you planning to add agency? " Yeah. So, we actually we do white label SEO for a lot of big SEO agencies. So, if you're interested in that, feel free to book a call via julian. com. If you're looking to supply us links, we're not really looking for like people to supply to buy links from. We actually we have approved partners already. We're not looking to hire anyone else for that. But if you want to hire us for doing link building, whether that's white label or for your own website, feel free to book in a call via julianggoldie. com.