Hermes Agent amb MiniMax M3: recerca i codi a baix cost

You can actually run Hermes agent for 100 times cheaper. Here's how. So, a major AI breakthrough just happened. For the first time ever, an AI model broke the price to capability line where the more you pay, the more you get. And everything so far has been either on this line or below. But Minimax M3 is a new AI model that for the first time ever is above this line. But most people, for some reason, still only use Claude or GPT, which is costing them hundreds of dollars per month every month. While Minimax is nearly free. Oh, and it's also really really good. On SWE Bench Pro, it outperforms not only Gemini 3.1 Pro, but also GPT 5.5. On SVG Bench, it outperforms Opus, GPT 5.5, and Gemini. And on BrowseComp, which is very important for Hermes agent, it is better than Opus and roughly similar as GPT 5.5. And this is just one of many reasons why Minimax M3 is perfect for Hermes agent. It's also cheap enough to run 24/7, which makes all ways on agents finally possible. Also, given the large context window of Minimax M3, it can do up to 1 million tokens, it can hold your entire project in memory, not forgetting anything. It's also built for long tool-heavy loops, such as the slash goal feature inside of Hermes. In fact, Minimax M3 can run for over 24 hours straight doing up to 2,000 different tool calls with no human intervention. That's incredible. Now, you might be thinking, "But David, how is all of this even possible? How come Minimax M3 can outperform models 10 to 50 times more expensive?" Well, the answer is the architecture. It has a novel architecture called MSA, which stands for Minimax sparse attention. And I know it sounds complex, it sounds difficult, but actually it's very simple and very easy to understand. And understanding it will give you an unfair advantage over everyone else. So, here is MSA clearly

explained. Normal transformers read every single token in context every time. That's why my context workflows cost a lot of money. MSA, on the other hand, checks which parts of the context matter and then only reads those parts, skipping everything else. So, the result is a full 1 million context window at 1/20th of the compute. And of course, the pricing of MiniMax is just insane. For comparison, Opus 4.8 costs $5 for a million input tokens and $25 for a million output tokens. MiniMax M3 costs $0.60 and $2.40. And on top of that, now it's 50% off, which makes it 10 to 20 times cheaper than Opus. This is how you can run Hermes agent 24/7 for just a couple of dollars. Real quick, are you building something impressive? I'm going to be hopping on a call personally with people who are building impressive stuff with AI. So, if that's you, make sure to fill this out. Again, it's the second link below the video. It's completely free to apply. Now, to put the insane cost-effectiveness of MiniMax into perspective, when you get the $20 a month plan, you get 1.7 billion tokens per month. Not million, billion. And at the current pricing, okay, not the discounted 50% off pricing, the full pricing, this plan, 1.7 billion tokens, gives you a blended average of 1,326 dollars per month of API usage, okay? Which, if we compare that to Opus 4.8 tokens, would be 11,900 dollars of monthly API credits if you get the $20 plan for MiniMax. I mean, for the max plan, which costs $50 a month, you literally get over 5 billion tokens per month, which I don't think you can get anywhere else. And since many of you are interested in running Hermes agent as cheaply and efficiently as possible, I reached out to the MiniMax team to sponsor this video and they agreed. So, if you want 12% off any of these paid plans, click the first

link below the video and make sure to apply. Huge thank you to the MiniMax team for sponsoring this video. With that being said, let's get to building. So, first, go to Google, while it still exists, and type in Hermes agent GitHub to go to the official GitHub for Hermes Agent. So, obviously Hermes is fully open source, so it's free to download. Scroll down here until you reach the quick install section, which gives you this one command. Click on that. Boom, open any terminal on your computer. I'm just using the default macOS one, but you can use C marks, T marks, whatever you want. Paste it in, hit enter, and this will begin installing Hermes Agent on your computer. If you don't have it installed, it will take like 2 minutes to install it. And then, we need to select the MiniMax token plan. I have the $20 a month, but I do recommend getting the $50 a month if you hit the limit. But, it's very hard to hit limits on 1.7 billion tokens. Yeah, the the value you get for this is really incredible, and Floppy cannot compete with this at all. Like, if you have the Claude $20 a month subscription, you're hitting limits constantly, every single day multiple times. With this, good luck. You'd have to be a serious AI engineer to hit limits, right? When the install finishes, you'll see the option to select the plans. So, all the way scroll down here, and you can see MiniMax. Here we have three options: MiniMax off, MiniMax default, and MiniMax China. So, here, what you want to do is select the first one, MiniMax, and we need to get the API key. So, again, when you click the first link below the video, you'll be brought to the MiniMax pricing plan. Here, you can go to the top right, click on console after you create an account, which is super easy. You can just create one with

Google. Takes like 10 seconds. So, in here, inside of the MiniMax platform, go to the left and click on plan details, and here you have your subscription key. This allows you to use your subscription to power Hermes Agent and many other AI agents, such as Open Code, which I'm going to show you later in the video, which actually you can use MiniMax M3 to build anything, build any type of software, whether it's a AI startup, whether it's internal software, whether it's just a cool demo for you, for your friends, for your family. You can use it inside of Open Code, which is also free and open source, with the subscription key from here. And again, I'm going to show you that later in the video, but first, copy it, switch back to the terminal, and paste that in. Okay, here it's asking for the base URL. You can just hit enter, and for the model, obviously select MiniMax M3. Okay, here you can keep current. messaging platforms. We can skip that for now and we can click done. And just like that, we've set up Hermes Agent locally. So, I can do clear and type in Hermes and this should launch the Hermes Agent TUI terminal user interface. Here we go. We have MiniMax M3 here. We can do hey and this is powered with my token plan, right? So, I'm using the $20 a month plus plan from MiniMax to power Hermes Agent and again, this gives me 1.7 billion tokens per month. So, you can literally use Hermes Agent 24/7 and you'll probably not run out. And if you do run out, you upgrade to the 50 plan which gives you 5.1 billion tokens and then you will not run out. So, what I'm going to do is I'm going to CD into a folder and I'm going to launch Hermes Agent in here and I'm going to tell it use the following

repository to save any and all files. Do not create or modify files outside of this project. Okay, send the delimiter and I'm going to send it the prompt. So, this is a deep research prompt and basically I'm going to compare it how good it would perform against something like Perplexity AI computer. So, I'm going to send the same prompt into Perplexity AI computer which of course cost $200 a month. Uh that's a lot more expensive and the usage is way less. Okay, so we can easily run out of these $200 a month these credits inside of Perplexity AI computer, but for MiniMax you can actually check your plan plan usage and you can see that so far we are at 1% of our weekly quota. Okay, so you can do many many deep researches with this and you're not going to hit your limit. So, this Hermes Agent will be running and doing this deep research about events in Katowice, Krakow and Warsaw over the next 90 days. So, different AI tech business events and I want to put everything into a single MD file. So, yeah, it's going to do this deep researches and I'm going to let it run. In the meantime, let me show you how to build anything with MiniMax M3 by using it inside of Open Code. So, if you don't know Open Code, it's like a open source alternative to closed code, and again, it's completely free. So, you can just go to open code.ai and copy this one-liner installer command. And while this is running, I'm going to do a new terminal, and I'm going to paste that in to install it locally on my MacBook. As you can see, it's pretty fast. Boom, there it is. So, let me do clear and type in open code. And of course, we want to use MiniMax inside of open code. So, type in {slash} connect to connect a new provider. I'm

going to type in MiniMax. There it is. minimax.io, and we need to put in the same API key. So, let's go back to plan details here and copy the subscription key. Boom. Paste that in and hit enter. Select MiniMax M3, and here here we are. Inside of open code, literally, how fast was that? Like 20 30 seconds? And we are chatting with MiniMax M3 inside of another agent, right? So, not only do we have Hermes agent here doing a deep research using the subscription. Here, we have open code also using the MiniMax token plan subscription, but here we can optimize it for coding. So, I'm going to give it a prompt. So, the first thing I want to test is a simple 2D game, Doodle Jump platformer game as a single contained HTML file. So, let's see how good it is at that very use case. And inside of open code, we can see the usage at the bottom. So, so far, we've spent 1 cent, 1 penny. We're going to see how expensive this is, but I already know that this is not even going to be comparable to Opus or GPT-5.5 because of how insanely efficient MiniMax M3 is. And let me highlight this once more because I think a lot of you don't realize the significance of this. So far, every single model has been on this line or below, which means you have to pay more to get more capability, but MiniMax M3 broke the line. It's the first model that gives you way more for the same amount, completely crashing Sonnet 4.6, by the way, which is not only over 10 times more expensive, but also worse at long horizon coding tasks, which is the type of stuff we're currently testing, right? And again, I'm going to be jumping on a call with a few of you who are building the most impressive things with AI completely for free. So, make sure to fill this out.

It's the second link below the video. All right, let's check on Hermes how it's doing. Still deep researching. So far used 13% of the context window, 60,000 tokens. Inside of Open Code, we are at still 1 penny. That's insane. Let's check the plan details here. Let me reload plan usage. Okay, so we're at 9% of our 5-hour limit and still 1% of our weekly limit. Incredible. I want to see how good this game will be. And actually, one more thing I want to test while these two are still running is how good Minimax M3 is at creating SVG files, right? And animations with SVG because it's crashing on the SVG bench eval. So, what I'm going to do is I'm going to open a new terminal, command N, type in open code. We can launch another Open Code instance while the previous one is running. And we can tell it build the following SVG animation. And I'm going to paste in a description of what I want to build. A single self-contained SVG file utilizing animate, animate transform, blah blah blah. It's a pretty long prompt describing exactly what I want. So, what this should show is a looping animation of different rifle parts assembling, shooting a shot, and then fading out. So, it's not just a simple image static image. It should be a SVG animation. And we can see that there's already thousands of tokens of Minimax. Let's check the usage. This is the main thing, guys, because I'm telling you you can still less than 1% of the weekly limit. Incredible. Let's look at the Hermes deep research. It's still researching. It's being very very thorough. Running for 12 minutes. I told you it can run for over 24 hours doing nearly 2,000 tool calls without any human interruption. And again, if you want 12% off any of the token plans from Minimax, make sure to click the first link below the video and get any of these plans for

yourself. You're going to get 12% off of this already incredible price. Okay, Okay, it seems like Hermes agent is finalizing the deep research with 20 different events happening over the coming weeks and months. And the huge benefit is that obviously Hermes Agent is open source running locally. Minimax is going to be open source. So far they open source previous models. And the cheapest plan is 10 times cheaper than the cheapest Perplexity computer plan, which obviously is $200 a month. And again, I'm running out of this usage pretty often. Like even for $200 a month, this doesn't last at all. With this, you'll probably get like 50x more deep researches inside of Hermes than with Perplexity computer. And again, this is cloud hosted. So who knows where these files are stored. You kind of have to trust Perplexity with all of this and all your data. Okay, there it is. So here is the formatting. Coverage by city Katowice 3, Krakow 7, Warsaw 10. I mean, yeah. Warsaw is the biggest city, so that makes sense. Top five local recommendations. Cybersecurity Expo and Forum Katowice June 15-16. Interesting. And here we have all the sources at the bottom. Very clean report, not overly bloated. That's also important. It's not like hundreds of pages of just AI slop. Let's look at what Perplexity computer gave us. A lot of these events are the same. I mean, you cannot really invent, you know, other events like that. Okay, so the events it recommended are different. Interesting. Actually, what we can do, I'm going to kill this Hermes Agent. Clear Hermes. And say, "List out all folders in this repo and list out all files in /reports folder." This is a separate Hermes, a separated session, so it will have no bias towards which is better. And we can have Minimax analyze these two reports and compare them. Okay, now analyze both of these markdown files, read them in full, and give me a

clear and objective comparison. Be very concise. All right, the SVG should be finishing up now. So, let's look at it. Comparison format. This is Perplexity versus Event Selection. This is a structure table, yeah? Event overlap. Okay. Unique to Perplexity Computer. So, some events are unique. Some of them overlap. Perplexity was stronger on small free meetups. Events is stronger on big conferences. The top five picks were different. And let's get the verdict. Events MD is the stronger calendar. Perplexity is better in niche meetups. Neither is complete. Merging them. So, that's crazy. The fact that these are comparable deep research reports where one of them is 10 times cheaper and gives you probably like my best estimate is like 40 to 100 times more deep researches per month compared to what you would get on the $200 month plan here. It's really insane the efficiency of the Min Max token plan. We're at 3% of our weekly limit and we're running multiple things here still. This SVG is going to be crazy. I don't know why it's cooking for so long, but this better be good. And look at the spend. Nine pennies. Nine cents. And let me remind you that this is not some cheap 7B model. In fact, Artificial Analysis puts it as one of the top models in terms of intelligence. Literally neck to neck with Gemini 3.1 from Google, which is way bigger model, way more expensive to run. And on the intelligence index, it literally is ranked as one of the best models in the world. So, this is not just some AI model, small model that's cheap to run. This is literally one of the best models out there that's also the most cost-efficient model to run. Okay, so we have this SVG file finally. Took a while because it kept overthinking. So, I told it to write the file. Let's see. Okay, great. It's written. So, I say, "Now, open this

SVG for me in Brave browser as a new tab." You can literally do anything if you just speak English. Like you need to verbalize your thoughts and describe what you want the agent to do and it can just do it. All right, so there it is. Rifle is being assembled. Okay, that's actually pretty good. Right, chambered, shot the gun. Now it should it disappear. This is impressive, guys. And again, $0.25 SVG animation that can be used in video games, in graphics, in videos, video editing, landing page on your site, and the shell is running out. Guys, this is This is good. This is good. For $0.25 pretty insane. Let's see the other um instance. This is building the 2D game. This is a bigger project, admittedly, but I also had to tell it to stop overthinking because this model is very perfectionist, right? I mean, there's a reason why it's performing really well on all these benchmarks. This model likes to think a lot. It likes to reason a lot. And sometimes that's good when you want a really detailed report, a really impressive refactor. But if you want something simple and quick, you need to kind of tell it like, "Stop planning, stop thinking. Do not overthink. Just write the file now." So, hopefully, we can test the game soon. So, the 5-hour limit, we're at 25% and by the way, this resets in 10 minutes, right? So, we cannot even get to How do you even use this out? Like, I'm using free agents, two open codes, one Hermes agent, and I'm even struggling to hit the limit. This is pretty impressive, okay? You literally need to be running four to eight agent in parallel 24/7 to even like hit these limits. Anyways, we already know it can do detailed deep research on the level of perplexity computer or even better while doing that for a fraction of the cost. We already know it's great at

SVGs. In fact, it outranks other frontier models at SVG bench. So, let's see how good it is at 2D games. Uh and yeah, this is the beautiful multimodality of Minimax and free. It's great at video, images, coding, text, all of the things you would expect a frontier AI model to do in 2026. And by the way, the Minimax team has committed to open weights. So, this model will be open weights or on Hugging Face and GitHub within 10 days of its release. So, that's going to be around June 10th, which means it will probably make MiniMax M3 the most powerful open weights model in the world the moment they open weights. It's not open source, but it's going to be open weights. And in fact, we have the HTML file done. So, I'm going to say open that in Brave browser as a new tab. And let's see how the game performs. It's 1,100 lines of code. And here we are. Hopper, okay. The graphics are nice. Okay, how do we play it? Resume. We need to jump. Okay. Look at the sounds. Yeah, I need to pause my music. Okay, the sounds are pretty solid. Is there some boost? No. What are the What are the keyboard controls? Answer in short. Okay, A and D and Okay, so the issue is some of these Look at this. Graphics are nice. The sounds are nice. There's no double jump, okay. Make sure the platforms are closer to each other. That is the main issue. Most of these levels have the platforms too far apart. Okay, this one should be playable. Look at this. Let's jump to the left. Okay, too far. Yeah, it needs to be either a double jump mechanic or the platforms need to be a bit closer. But not a bad game. Certainly not a bad graphics. And again, open weights model, all available for like what? 0.4 Pretty crazy. All right, let's let's reload this

actually. The gap should be smaller. All right, here we go. Let's see how it goes. What is this? Some boost? Oh, it crashed. I'm going to send it a screenshot and obviously we can fix it and keep improving it forever. This, uh, type of platform crashes the game. It's going to analyze that image and, you know, fix the bug. You get the point. So, you can build any type of software with it, whether it's a 2D game, whether it's SVG graphics, whether it's deep research, full-stack web app, internal software. This model is inherently multimodal and it's competent on all of the frontier and agentic capabilities. It can basically do anything you need it to do for a fraction of the cost that other models other models are. So, again, most people are stuck to just using ChatGPT and Claude. If you sit down and if you actually watch this video again and implement MiniMax, you're going to be able to do everything these models can do for like 10 to 20 times less. So, if you care about cost and if you want to run Hermes Agent in the most efficient way possible, set up the MiniMax token plan, connect it like I did in the video, and again, it's the first link below the video. Just choose the plan that works for you. If your budget is limited, go with the $20 a month. If you want 5 billion tokens per month, go with the max plan. But, either way, click the first link below the video to get 12% off. And with that being said, thank you guys for watching and have a wonderful rest of the week.